CAERDROIA 50 - - [PDF Document] (2024)

: L :



The Journal of Mazes & Labyrinths



The tomb of the Marchioness of Waterford in the churchyard

of St. Michael & All Angels, Ford, Northumberland, England, ca. 1891.

Photo Jeff Saward, July 2021


The Journal of Mazes & Labyrinths


Cover : The Walls of Troy, Marfleet, England; from Ackermann’s Repository of Arts, 1815, original

engraving in the Labyrinthos Archive

1 Frontis : The tomb of the Marchioness of Waterford, ca. 1891; Photo: Jeff Saward, July 2021

3 Editorial : a note from the editor, Jeff Saward

4 New Discovery of Stone Labyrinths in Western Maharashtra, India : Sachin Bhagwan Patil

& P.D. Sabale report important new discoveries from India

11 Keeping Kern Current: Locating ‘Lost’ Labyrinths in Medieval Manuscripts : Jill K.H.

Geoffrion & Alain Pierre Louët search the library archives

26 Hedge Mazes in Portugal: A Brief History : Carlos Soreto presents a history of Portuguese

mazes and details two particular examples in Oporto

32 The Surroundings of the Rösaring Labyrinth : Heather Robertson describes this important

Swedish stone labyrinth and records the work of its official guide, Börje Sandén, who died


38 Historic Turf Maze Sites in Wales : Jonathan Mullard looks for possible sites of turf

labyrinths now long gone

44 From Jerusalem to Troyborg: The Labyrinth Name Change of the North : Christina

fa*gerström explores the origins of two popular labyrinth names

51 Simple Alternating Transit Mazes : Richard Myers Shelton studies the structure of classical

and Roman labyrinths and records the work of the late Wiktor Daszewski

67 A Mysterious Medieval Maiden : Jill K.H. Geoffrion & Alain Pierre Louët examine an

unusual medieval manuscript labyrinth and its surprising inhabitant

69 The Minnie’s Gap Labyrinths : Kirk Astroth reports on two labyrinths from Wyoming, USA

71 Notes & Queries : an inscribed powder horn from the American War of Independence, a

labyrinth in a temple in Myanmar, a labyrinth on an 18th

century English sampler and the

“Modern Labyrinth” ca. 1900

73 Caerdroia : submission details, subscriptions, etc.

74 Labyrinthos : who we are and what we do, etc.

Caerdroia 50 was produced during Spring 2021 by Jeff and Kimberly Saward at Labyrinthos HQ.

Opinions stated by contributors are not always those of the editors, but Caerdroia welcomes open

discussion and endeavours to provide a forum for all who are lured by the labyrinth.

Editor & Publisher: Jeff Saward – Associate Editor: Kimberly Lowelle Saward, Ph.D.

Caerdroia 50 is © Labyrinthos/individual authors 2021, as appropriate.

Caerdroia 51 is due for publication Spring 2022, submissions by December 2021 please.


Editorial – Caerdroia 50

Jeff Saward, Thundersley, July 2021

Welcome to the 50th

edition of Caerdroia, a little later than planned, due to various other

commitments, but now finally published and bringing the history of Caerdroia to the half-

century mark, and our largest edition so far. With the Covid-19 pandemic still hampering

trade and travel, this 50th

edition has again been published in digital PDF format initially,

with the printed version to follow in due course.

Despite the disruption, this edition contains articles on mazes and labyrinths from various

corners of the world – four recently discovered stone labyrinths in Western India, a

catalogue of medieval manuscript labyrinths not documented in Hermann Kern’s Through

the Labyrinth and a look at one especially unusual example, historic hedge mazes in

Portugal, the Rösaring stone labyrinth in Sweden, possible turf maze sites in Wales, the

names of labyrinths in northern Europe, the structure of classical and Roman labyrinths,

two labyrinths in Wyoming, one in Myanmar, another on a stitched sampler from England

and a fascinating example on a powder horn from the American War of Independence – as

always a packed and diverse edition. Two of the articles this time around also remember

researchers sadly no longer with us, Börje Sandén and Wiktor Daszewski – a reminder,

should it be needed, of the importance of recording our research and ideas for future

workers in the field.

The Labyrinthos and Caerdroia website – – now contains a wealth of

material, including a wide selection of downloadable PDF files of commonly requested

articles from old out-of-print editions of Caerdroia, digital editions of Caerdroia 33 through

39, and selected articles from more recent editions, along with some of the many photos and

graphics in our extensive photo library and archive. This edition of Caerdroia is also

available on the website for subscribers at:

Our next edition, Caerdroia 51, is scheduled for publication in spring 2022. As always, if you

have a paper or shorter article you wish to submit for inclusion in the next edition, send it

to me as soon as possible, along with the usual labyrinthine snippets and curios that help fill

the pages.

Jeff Saward – e-mail: [emailprotected] – website:



New Discovery of Stone Labyrinths in Western Maharashtra, India

Sachin Bhagwan Patil & P.D. Sabale


Stone labyrinths have recently been discovered at four new localities in parts of the Sangli

District of Maharashtra in the western-central region of India. Three, situated in Walwa

Tehsil and a fourth in Kavtemahankal Tehsil, throw light on an ancient trade route between

the early historic settlements and trade centres of Kolhapur and Karad, as well as between

Kolhapur and Ter. These suitable routes are discussed in cultural, historical, archaeological,

geological and geomorphologic points of view.

The authors argue that these labyrinths, in the south-central Deccan plateau region just

after the Sahyadri escarpment, were constructed alongside ancient trade routes in the early

historic period, and were used by travellers following them and visiting the Buddhist caves

along the roads. The stone labyrinths were constructed at the junctions and passes along

these ways.


The labyrinth is one of the oldest contemplative devices known to humankind, and dates

back more than 3000 years. In India, the labyrinth has been recorded from Harappan to

colonial periods in different forms, from many different locations. These labyrinths are

depicted as rock paintings and carvings, as images and pavements in temples, and

constructed of stones laid on the ground.

Behind the design, drawing, and construction of these labyrinths lie hidden meanings which

vary from place to place. Most of the time, they are connected to the history of that area

and well-known stories from Indian literature and mythical epics. The labyrinth was used to

illustrate the strategic formation used in warfare known as chakravyuh; in ancient epics as

manaschakra, a symbol used in personal and spiritual growth; as rangoli, a symbol used in

traditional folk art and in mythology as yamadwara. It may also have served as a mark of

significance for confusing places and passes on ancient trade routes and the stone labyrinths

may be seen as a landmark in these places.

Study Areas

The study areas of Aitawade Budrukh and Vashi are situated in the middle reaches of the

Warana river basin and politically in Walwa Tehsil of the Sangli district of Maharashtra,

west-central India.


i. The Aitawade Budrukh Labyrinths

The first labyrinth, 11.7 metres in diameter and of 7-circuit classical form, is situated 2.2

kilometres northeast of Dhagewadi, the nearest settlement, on rocky ground in the hilly,

scrub forest of the Peth Forest Reserve, on right side of the cart road, near the Khandoba

Temple that is located around 25 meters northwest of the labyrinth. Latitude: 16.993755 N,

Longitude: 74.206457 E.

Fig. 1: The first labyrinth (Aitawade Budrukh 1) 2.2 km NE of Dhagewadi

and a plan view of the labyrinth

The second example, 9.2 metres in diameter and also of 7-circuit classical form, is located

1.75 km NE of Dhagewadi and 0.5 km south of the previous labyrinth, in the Aitawade

Budrukh reserve forest area, along the south side of the bullock cart road near the Pavatka

temple that is located around 20 metres southwest of the labyrinth. Latitude: 16.987677 N,

Longitude: 74.207794 E.

Fig. 2: The second labyrinth (Aitawade Budrukh 2) 1.75 km NE of Dhagewadi

and a plan view of the labyrinth


ii. The Vashi Labyrinth

The third labyrinth is located 3.0 km east of the first two labyrinths, 1.1 km SSE of the village

of Shivpuri, the nearest settlement, and stands on a rocky surface in a pass in between two

hillocks, near the boundary of the Kameri forest reserve. This location is at the remote

northern boundary of the Vashi village landholding. The outermost circuits of this labyrinth

are somewhat damaged, but the design was originally of 7-circuit classical form. Its current

diameter is 10.8 metres, but it would originally have been around 13.5 metres. Latitude:

16.988316 N, Longitude: 74.236290 E.

Fig. 3: The third

labyrinth (Vashi) 1.1

km SSE of Shivpuri,

and plan views of the

labyrinth as currently

preserved and with the

missing circuits


This area is bounded by Retharedharan (to the NW); Shivpuri (N); Kameri (NE);

Jakraiwadi (E); Ladegaon (SSE); Aitawade Budrukh (S); Dhagewadi and Karve (SSW). It

is mapped on the Survey of India toposheet map 47 L/1, 1:50,000 scale.


iii. The Malangaon Labyrinth

The other study area of Manerajuri is situated in the Agrani river basin and politically in

Kavathemahankal Tehsil of the Sangli district of Maharashtra, west-central India, just over

50 km to the east of the other three labyrinths described above.

The fourth labyrinth is situated on a rocky hillock area 4.5 km southwest of Malangaon

village, in the area known as Kodyacha Maal, on the north side of the express roadway

between Sangli and Solapur, via Kumathe, and under the jurisdiction of Manerajuri, the

nearest large village, to the west. The shape and form of this labyrinth is slightly irregular,

but is clearly of 7-circuit classical form and 9.9. metres wide. A modern enclosure has been

built around the labyrinth to protect it. Latitude: 17.020806 N, Longitude: 74.747338 E.

Fig. 4: The fourth labyrinth (Malangaon) 4.5

km SW of Malangaon,

and a plan view of the labyrinth

This area is bounded by Bhosalenagar (to

the NW) and Manerajuri (W); Yogewadi

(SW); Boargaon (E); Malangaon and

Gavahan (NE).


Geology and Geomorphology

The geology of the study area of the Aitawade Budrukh and Vashi labyrinths is a low

elevation region with circumdenuded and isolated mesa-type features typical of the Deccan

volcanic land present on the plateau region of Maharashtra, close to the foothill region of

the Sahyadri escarpment. This moderate to deeply weathered hilly landscape with scanty

vegetation of thorny plants and seasonal grasses is the source of the Mallikarjun hill stream,

a minor tributary of the Warana river in its upper reaches. This hill also acts as a ridgeline

separating the Warana basin on its right side and Krishna River on its left. Geologically, the

study area belongs to the Southern Deccan Volcanic Province of the Deccan Traps

formation of India. It is one of the largest volcanic provinces on the earth and was formed

during the upper Cretaceous to lower Eocene period as a result of fissure type lava eruptions

[Sabale 2008].

The Aitawade Budrukh labyrinths are located on sub-hillock plateaus aligned on the same

hills, but on opposite sides of a valley divided by a stream, and along this gully a bullock cart

road is present. This road was considered an important junction on an ancient short-cut

route, which leads from the Konkan coast of Maharashtra and Goa to the west, linking

Kolhapur and Karad.

Fig. 5: Ancient trade routes in Maharashtra and the locations of the labyrinths


Ancient trade routes and trading centres

During the Satavahana dynasty (late 2nd

century BCE to early 3rd

century CE) control of trade

and points of exchange was a main source of revenue. The rulers were responsible for

offering protection to the caravans on their journey and were entitled to a fee for the service

[Gurukkal, p.264]. This dynasty owed their prosperity to trade, a view justifiably supported

by most historians, according to whom the period witnessed a remarkable growth of inland

as well as overseas commerce in the Deccan areas. The Satavahana country was well known

to the Greco-Roman geographers and navigators as a landscape full of forest goods and the

Deccan Plateau was rich in forested mountain and wild animals. Consequently, there were

several inland points of exchange acting as feeders for the Satavahana ports along west coast

of India. The Greco-Roman merchant mariners were visiting these ports and bartering

metals including tin and Italian bronze artifacts and also fine quality coral, as demonstrated

by archaeological finds at the domestic site of Kolhapur, which is close to the study area.

Karad and Kolhapur were important trade centres at this time, and the port of Dabhol-

Khed connected to the Karad route through the Hatlot Ghat, so called because the slope of

the ghat was so steep and difficult that the drivers of loaded carts had to get down to push

the carts uphill (hatlot meaning hand pushing) [Hebalkar, p.27]. At Brahmapuri, located in

vicinity of Kolhapur city, three bronze mirrors were excavated, one an authentic Roman

import and two that may be Satavahana copies, along with a figure of Poseidon, the Graeco-

Roman god of the sea and a clay bulla prepared from a Roman coin [Sankalia, p.9 & 97].

Similarly, two heavily corroded bronze mirrors, iron implements, bangles, Roman pottery,

Satavahana and Roman coins were reported from another Satavahana site at Ter, lying to

the northeast of Kolhapur [Suresh, p.127].

Discussion and Conclusion

To understand the relationship of these labyrinths with the surrounding culture, a detailed

site catchment analysis was carried out. Through a systematic village to village survey in and

around these locations, we explored a number of sites of different cultural periods. While

further research work is required to understand the importance and function of these

labyrinths in Maharashtra, as discussed above, the study area of the Aitawade Budrukh and

Malangaon labyrinths is at a point where two ancient trade routes from Kolhapur divide,

i.e., the one going to Karad and another towards Ter. The location of the labyrinths at

Aitawade Budrukh was at a major junction below the upper shield land, on a medium

elevated and dissected ridgeline. From this ideal location, both the sites are visible,

accessible by good road connection, and nearly equidistant. Therefore, after looking at such

suitable landform characters, it seems they may have selected this location for the

establishment of these labyrinths. Likewise, the labyrinth at Malangaon is situated alongside

the road linking Kolhapur and Pandharpur. As discussed above, goods from the Konkan

ports were loaded on bullock carts and carried through the ghat section of the plateau region

to different localities in various directions. Therefore, new travellers can select the proper

road to reach their destination and the labyrinths at this junction give proper direction to

the travellers to reach their destination. In this process, these four labyrinths in Maharashtra

all played a very important role in guiding travellers along these roads. These newly

described labyrinths will prove to be important evidence of ancient Indian culture and trade

relations. Further research work is required to understand the importance and function of

the labyrinths discovered in the region of Maharashtra.


Sachin Bhagwan Patil and P.D. Sabale, Department of A.I.H.C. and Archaeology,

Deccan College Deemed University, Pune-411 006, Maharashtra, India; April 2021

Email: [emailprotected]


The authors are thankful to Prof. Dr. Vasant S. Shinde, Director General, Maritime Heritage

Complex Gujarat, former Vice Chancellor of Deccan College Deemed University, Pune, for his

guidance and encouragement during progress of the work. The authors would like to express our

deepest gratitude to Jeff and Kimberly Saward of Labyrinthos. The authors are also thankful to

Mr. Shahaji Gaikwad and Mr. Tukaram Gaikwad of Aitawade Budrukh, Mr. Amrut Patil of Vashi

and Mr. Ashish Shinde Sarkar of Malangaon for their help during the area exploration and also

kind gratitude for to giving their time. Similarly, we are thankful to Mr. Arvind Asabe, Dr. Shivaji

Shirsagar, Mr. Sagar Patil, Miss Komal Patil and Miss Meghana Agarwal of Deccan Collage Pune,

for help in reconstructing the material and history of the study area. We offer thanks to Mr.

Rajkumarji of WIELDY, Mumbai for providing maps & GIS designs of locations and Mr. Anil

Kisan Dudhane and Miss Sayali Ranavade of the Marathidesha Foundation, Pune, for digital

sketches. Finally, thanks go to Mr. Aman Sinha of Kolhapur for aerial images.


Gurukkal, Rajan. Rethinking Classical Indo-Roman Trade: Political Economy of Eastern

Mediterranean Exchange Relations. London: Oxford Press, 2016.

Hebalkar, Sharad. Ancient Indian Ports with special reference to Maharashtra. New Delhi:

Munshiram Manoharal Publishers, 2001.

Kern, H. Through the Labyrinth. Munich: Prestel, 2000. See chapter XVII, p. 284-297.

Kraft, J. “The Oldest Labyrinth in India?” Caerdroia 35 (2005), p. 57-59.

Kumar, A. “Labyrinths in Rock Art: Morphology and Meaning with Special Reference to India.”

Heritage: Journal of Multidisciplinary Studies in Archaeology 3 (2015), p. 84-106.

Kumar, A. & S.K. Tiwary. “Two Interesting Labyrinth depiction from Bihar.” Journal of the

Indian Archaeological Society 44 (2014), p. 276-280.

Kürvers, K. “Kota Labyrinths in Southern India.” Caerdroia 36 (2006), p. 38-52.

Murugan, S. “A Newly Discovered Stone Labyrinth in India.” Caerdroia 44 (2015), p. 56.

Murugan, S. “More Labyrinths in Tamil Nadu, India.” Caerdroia 45 (2016), p. 52.

Sabale, P.D. “Morphosectonic Studies of Deccan volcanic plateau from Sangola gravity high.”

Unpublished thesis, Shivaji University, Kolhpur, India, 2008.

Sankalia, H.D. Excavations at Brahmapuri (Kolhapur) 1945-46. Pune: Deccan College Post

Graduate and Research Institute, 1952.

Suresh, S. Symbols of Trade: Roman and Pseudo-Roman objects found in India. New Delhi:

Manohar Publishers, 2004.

Saward, J. Labyrinths & Mazes. London & New York: Gaia/Lark, 2003, p. 60-66.

Saward, J. “A Labyrinth at Kurukshetra, India.” Caerdroia 35 (2005), p. 59.

Saward, J. & K. “Labyrinths in Western India.” Caerdroia 36 (2006), p. 59-62.

Schuster, C. Social Symbolism in Ancient & Tribal Art. Ed. E. Carpenter, NewYork: Rock

Foundation, 1988, vol. 3:2, p. 288-299.

Ward, Philip. Western India: Karnataka, Bombay, Maharashtra. Cambridge, UK: The Oleander

Press, 1991.



Keeping Kern Current: Locating ‘Lost’ Labyrinths in Medieval Manuscripts

Jill K. H. Geoffrion & Alain Pierre Louët


First published in 1982, Hermann Kern’s Labyrinthe,

the catalogue of labyrinths which has become

indispensable to scholars, devoted chapter seven to

labyrinths found in manuscripts and noted 80 medieval

examples. In 2000 an updated English language version

of Kern’s book, Through the Labyrinth, was edited by

Jeff Saward and Robert Ferré, in which they included

an additional four manuscript labyrinths. Since that

time, many other manuscript labyrinths have been

identified by scholars and others have been put on-line,

either as part of library digital collections or used as

illustrations in books or online articles.

Due to our interest in these labyrinths, we have searched out new examples both using the

internet and in-person when possible. We now offer the fruit of these labours, the 38

labyrinths below, with the hope that others will be able to use these examples in their work

and research. Whenever possible, we have included images of these labyrinths and links

where more information about them can be found.

Below we present these 38 manuscript labyrinths, in approximate chronological sequence,

along with a table of their essential details. A note on the terminology regarding the

labyrinth designs is in order at this point. In Kern’s original table (2000 edition, pages 107-

109) he basically uses six categories to cover the majority of the labyrinths – Cretan, Cretan

modified, Otfrid, Jericho, Chartres and Chartres modified – plus a few that are correctly

described as mazes. We have chosen to retain this basic system to allow direct comparison

of the entries that follow with Kern’s original catalogue.

Modern readers might be more familiar with the terms “Classical” instead of “Cretan,” and

“Medieval” rather than “Chartres,” but we will leave that translation of terminology to the

reader. We have added an occasional comment where Kern’s system falls short of adequate

description and have been consistent regarding the use of the term “Chartres-style” for

those medieval labyrinths that have the same path arrangement as the labyrinth in Chartres

Cathedral, and “modified Chartres” for those that have different path arrangements,

likewise for “Cretan” and “modified Cretan.” The number of circuits refers to the

concentric paths arranged around the central goal.


Table of Manuscripts

No. Collection Folio no. Labyrinth type Date

1 Paris Lat. 12048, fol. 80 Cretan 780-800 CE

2 Paris Lat. 4416, fol. 35 Cretan 9th


3 Paris Lat. 3840, fol. 1r unknown 9th


4 Berlin Lat. 356, fol. 11v Cretan 9th



5 Basel AN IV 11, fol. 77r Otfrid-style 11th


6 Orléans Ms. 16, fol. 252 Chartres-style 11th



7 Leiden BPL 92A, fol.182 Chartres-style 12th


8 Paris Lat. 5371, fol.240v Chartres-style 12th


9 London Cot. MS Tib. BII, fol. 248v modified Chartres 1110

10 St. Omer Ms. 684, fol. 74 Chartres-style 1120-11140

11 Cambridge Ms. H.11, fol. 124v Chartres-style 1180-1200

12 Paris Ms. 711, fol. C Chartres-style 12th



13 Leiden Ms. 100A, fol. 1 Chartres-style 1150-1200

14 London Add. 15603, fol. 142v unusual 1175-1200

15 Chantilly Ms. 0328, fol. 080v Chartres-style 13th


16 Geneva Ms. Gr. 44, p. 674 partial, uncertain 13th


17 Munich Clm. 17403, fol. 13 Chartres-style 1241

18 Cambridge Ms. 0.2.45, p. 001 Chartres-style 1248-1300

19 Paris Lat. 2809, fol. 153r modified Chartres 1270-1330

20 Amsterdam Hs. Ros. 609, fol. 127v modified Chartres 1290

21 Cambridge Ms. 0.2.5, fol. 27r Chartres-style 14th


22 Paris Ms. 8530, fol. 175r modified Chartres 14th


23 Paris Ms. Or. Heb. 9, fol. 236v Cretan 1304

24 Rome Ms. Or. 72, fol. 6v Cretan 1326

25 Berlin Hs. or. 2371, fol. 167v modified Cretan 1350

26 Berkeley US-BE m 744, fol. 31v modified Chartres 1375

27 Hanover, NH Taj Torah, 290 Jericho 1400-1450

28 Paris Greek 2055, fol. 53v Chartres-style 15th


29 Paris FR 17001, fol. 2v complex 15th


30 Paris FR 17001, fol. 27r complex 15th


31 Paris FR 17001, fol. 88r complex 15th


32 Philadelphia LJS 226, note 1, side 2 Chartres-style 1410

33 Paris Or. Per. 62, fol. 322v Jericho 1410

34 Dublin Per. 322, fol. 121r Jericho 1420

35 Den Haag KB 72 A 23, fol. 21v modified Chartres 1460

36 Princeton Ms. 158, fol. 157v modified Chartres 1471

37 Heidelberg Cod. Heid. Or. 118, fol. 197v Jericho 1475

38 Den Haag KB 128 C 4, fol. 40r modified Chartres 1512


The Manuscripts

1 - Paris, France, Bibliothèque Nationale de France; Lat. 12048, fol. 80 (780-800 CE)

This 7-circuit Cretan labyrinth is in the form of an

illuminated capital in a Gelasian Sacramentary,

the so-called Gellone Sacramentary, a book of

Christian liturgy related to the sacraments from

the Diocese of Cambrai or Diocese of Meaux in


1 - Lat. 12048, fol. 80; courtesy of

Bibliothèque nationale de France



2 - Paris, France, Bibliothèque Nationale de France; Lat. 4416, fol. 35 (801-900)

This 7-circuit Cretan labyrinth has a doorway at

the entrance and is found in the Epitome Aegidii of

the Lex Romana, a Latin legal treatise, from


2 - Lat. 4416, fol. 35; courtesy of

Bibliothèque nationale de France


General Information on manuscript:


3 - Paris, France, Bibliothèque Nationale de France; Lat. 3840, fol. 1r (801-900)

This labyrinth, identified in the catalogue of the Bibliothèque Nationale de France, has an

unknown pattern and exists in the Collectio Canonum Dionysio-Hadriana, a book of canon

law, from Saint-Martin of Spanheim, Mainz, Germany. There is no image of this manuscript

currently available online.

Description of the manuscript:

No image currently available


4 - Berlin, Germany, Staatsbibliothek;

Ms. theol. Lat. 356, fol. IIv (801-1000)

This 7-circuit Cretan labyrinth was placed on the top of a page

in book one of Homiliae in Ezechielem, a series of homilies on

the Hebrew Testament book of Ezekiel by Pope Gregory I,

from Werden in Germany.

4 - Ms. theol. Lat. 356, fol. IIv;

courtesy of Staatsbibliothek PK zu Berlin

Labyrinth: https://digital.staatsbibliothek-


5 - Basel, Switzerland, Universitätsbibliothek;

AN IV 11, fol. 77r (1001-1100)

This 11-circuit Otfrid-type labyrinth with long,

sweeping pathways is in a manuscript relating to the

De coniuratione Catilinae and De bello Iugurthino, the

best-known works of Gaius Sallustius Crispus

(Salluste), a Roman politician and historian from the

first century BCE. From the Carthusian monastery of

Chartreuse de Bâle, Basel, Switzerland.

5 - AN IV 11, fol. 77r; courtesy of

Universitätsbibliothek Basel



6 - Orléans, France, Bibliothèque Municipale;

Ms. 0016, fol. 252 (1001-1200 for the end matter

including the labyrinth, 901-1000 for the


This 11-circuit Chartres-style labyrinth is found at the

end in a manuscript of biblical texts from the Hebrew

Testament, books of Proverbs, Song of Solomon, Job,

Maccabees, and Tobias, from Fleury Abbey, Saint-

Benoît-sur-Loire, France.

6 - Ms. 0016, fol. 252; courtesy of

Bibliothèque Municipale, Orléans





7 - Leiden, Netherlands, Universiteit Leiden;

BPL 92 A, fol. 182 (1101-1200)

This 11-circuit Chartres-style labyrinth is found in a

manuscript of Octaviani Caesaris Augusti Versus in P.

Virgilii Maronis, a collection of Roman writings which

include poetry and grammar by Donatus, Servius

Honoratus and Virgil, from Normandy in France.

7 - BPL 92 A, fol. 182; courtesy of Universiteit Leiden

Description of manuscript:




8 - Paris, France, Bibliothèque Nationale de France;

Latin 5371, fol. 240v (1101-1200)

This 11-circuit Chartres-style labyrinth is found in a

collection of texts written by the Christian authors

Hincmarus Remensis, Eugene of Toledo and the

Venerable Bede, from the Abbey of Notre Dame, Mouzon

(Ardennes) in France.

8 - Lat. 5371, fol. 240v; courtesy of

Bibliothèque nationale de France


9 - London, England, British Library;

Cotton MS Tiberius BII, fol. 248v (1110)

This 18-circuit modified Chartres labyrinth with a unique

pattern is found in a manuscript on the martyrdom and

miracles of St. Edmund (king of East Anglia from about

855 until 869) by Abbo of Fleury, from Ely in England.

9 - Cotton MS Tiberius BII, fol. 248v;

courtesy of the British Library





10 - St. Omer, France, Bibliothèque Municipale;

Ms. 0684, fol. 74 (1120-1140)

This 11-circuit Chartres-style labyrinth is found in the

Consolation of Philosophy by Roman statesman

Boethius, from the Abbey of St. Bertin, St. Omer in


10 - Ms.684, fol. 74; courtesy of

Bibliothèque Municipale, St. Omer

Ms. including labyrinth:


11 - Cambridge, England, St. John’s College,

University of Cambridge;

Ms. H.11, fol. 124v (1180-1200)

This 11-circuit Chartres-style labyrinth is found in

Isidore of Seville’s Etymologies, from Wigmore Priory,

Herefordshire, England.

11 - Ms. H.11, fol. 124v; courtesy of

St. John’s College, University of Cambridge





12 - Paris, France, Bibliothèque de l’Arsenal; Ms. 711, fol. C (1101-1300)

This 11-circuit Chartres-style labyrinth is found in a

Miscellany Collection: Pierre Hélie, Quintilian,

Cicero, Seneca, Plato, Aulu-Gelle, Cassiodorus,

Celsus (Julius), Macrobe, Mela (Pomponius),

Petrone, Plautus, Sallust, Sidoine Apollinaire,

Suetonius, Terence, Varron, and Boèce, with texts

written in Latin and French, from the Abbey of St.

Victor, Paris, France.

12 - Ms. 711, fol. C; courtesy of

Bibliothèque nationale de France

Manuscript, with link to page with labyrinth:



13 - Leiden, Netherlands, Universiteit Leiden;

Ms. 100 A, fol. 1 (1150-1200)

This 11-circuit Chartres-style labyrinth is found in a

manuscript of the biblical book of Job, Liber Hiob ex

Hieronymi versione cum Genealogia et commentario,

which includes Jerome’s genealogy and commentary,

from the Netherlands.

13 - Ms. 100 A, fol. 1;

courtesy of Universiteit Leiden

Manuscript and labyrinth:


14 - London, England, British Library;

Additional 15603, fol. 142v (1175-1200)

This unusual 5-circuit, 4-quadrant labyrinth is found in

a manuscript copy of the Etymologies by Isidore of

Seville, from Neuvelle-lès-la-Charité in France.

14 - Additional 15603, fol. 142v;

courtesy of the British Library

Manuscript and labyrinth:


15 - Chantilly, France, Musée Condé;

Ms. 0328, fol. 80v (1201-1300)

This 11-circuit Chartres-style labyrinth with a woman’s

head at the centre, is found in the Petrus Hispanus

Thesaurus Pauperum, a collection of writings,

remedies, recipes, prayers and hymns to the Virgin,

from Spain.

15 - Ms. 0328, fol. 80v;

courtesy of Musée Condé, Chantilly

Manuscript and labyrinth:



16 - Geneva, Switzerland, Bibliothèque de Genève;

Ms. Gr. 44, p.674 (1201-1300)

This partial labyrinth of simple, but uncertain design, is found in

a manuscript copy Homer’s Iliad, from Constantinople (Istanbul),


16 - Ms. Gr. 44, p. 674;

courtesy of Bibliothèque de Genève

Manuscript and labyrinth:

17 - Munich, Germany, Bayerische Staatsbibliothek;

Clm. 17403, fol. 13 (1241)

This 11-circuit Chartres-style labyrinth is found on

the first page of the Glossarium Salomonis sive Mater

verborum. Herbae pictae cum explicatione, a glossary

that includes illustrated herbs, from the monastery

of Scheyern, Germany.

17 - Clm. 17403, fol. 13;

courtesy of Bayerische Staatsbibliothek

Labyrinth and manuscript: https://daten.digitale-


18 - Cambridge, England, Trinity College;

Ms. 0.2.45, page 001 (1248-1300)

This 11-circuit Chartres-style labyrinth is found in a

bound miscellany of various manuscripts that

includes music and writings in Latin, French and

Middle English, from Cerne Abbey, England.

18 - Ms. 0.2.45, page 001;

courtesy of Trinity College, Cambridge

Manuscript and labyrinth:



19 - Paris, France, Bibliothèque Nationale de France;

Latin 2809, fol. 153r (1270-1330)

This 11-circuit modified Chartres labyrinth (similar to

the design of the Sens labyrinth) is found in a

manuscript of Gregorius Magnus, Dialogi; Sententiae

Patrum a text that includes Gregory the Great’s

Dialogues and Tyrannus Rufus’s treatise on monks

and the Holy Fathers, from Narbonne, France.

19 - Latin 2809, fol. 153r; courtesy of

Bibliothèque Nationale de France



20 - Amsterdam, Netherlands, Bibliotheca

Rosenthaliana, Amsterdam University Library;

Hs. Ros. 609, fol. 127v (1290)

This 11-circuit modified Chartres labyrinth is placed in

the Esslingen Mahzor, a festival prayer book for Yom

Kippur and Sukkot, from Germany.

20 - Hs. Ros. 609, fol. 127v;

courtesy of Bibliotheca Rosenthaliana,

Amsterdam University Library

Information on the manuscript:


21 - Cambridge, England, Trinity College; Ms. 0.2.5,

fol. 27r (1301-1400)

This 11-circuit Chartres-style labyrinth is found in a

miscellany with canonical, astronomical texts and

medical treatises, of unknown provenance.

21 - Ms. 0.2.5, fol. 27r;

courtesy of Trinity College, Cambridge

Labyrinth and manuscript: https://mss-




22 - Paris, France, Bibliothèque de l’Arsenal;

Ms. 8530, fol. 175r (1301-1400)

This 11-circuit, Chartres modified labyrinth (with

several errors) is found in an Italian manuscript of

Dante’s Divine Comedy.

22 - Ms. 8530, fol. 175r; courtesy of

Bibliothèque Nationale de France

Labyrinth and manuscript:

23 - Paris, France, Bibliothèque Nationale de France;

Ms. Orientaux. Hebreu 9, fol. 236v (1304)

This 7-circuit classical rectangular text labyrinth, with

corner embellishments, is found in a Hebrew language

manuscript of the Hebrew Bible, from Germany.

23 - Ms. Orientaux. Hebreu 9, fol. 236v;

courtesy of Bibliothèque Nationale de France


24 - Rome, Italy, Biblioteca Angelica; Ms. Or. 72, fol. 6v (1326)

This 11-circuit Cretan labyrinth, depicting Jericho, is

in a manuscript of the biblical prophets, opposite the

initial folio of the biblical book of Joshua, from

Frascati, Italy.

24 - Ms. Or. 72, fol. 6; courtesy of

the Ministry for Arts and Culture, Biblioteca Angelica

Labyrinth illustrated:



25 - Berlin, Germany, Staatsbibliothek zu Berlin - Preußischer Kulturbesitz;

Hs. Or. 2371, fol. 167v (1350)

This 8-circuit Jericho labyrinth is found in an Iranian

manuscript of Muǧmal at-tawārīḫ wa-'l-qisas, a

compendium of histories and stories.

25 – Berlin; Hs. or. 2371, fol. 167v;

courtesy of Staatsbibliothek PK zu Berlin

Labyrinth and manuscript:


26 - Berkeley, California, USA, University Library;

US-BE m 744, fol. 31v (1375)

This 11-circuit modified Chartres labyrinth which

includes musical notations set on its paths is found in

a manuscript of the Treatises of Arezzo, Guido and

Goscalcus, from Italy.

26 - US-BE m 744, fol. 31v;

courtesy of Berkeley University Library

Labyrinth on YouTube:

Description of manuscript:

27 - Hanover, New Hampshire, USA, Rauner Library,

Dartmouth College; Ms. 290 (1400-1450)

This 6-circuit Jericho labyrinth is located in a copy of

the Taj Torah, from Yemen.

27 - Taj Torah, Ms.290;

courtesy of Dartmouth Library

Labyrinth and manuscript:



28 - Paris, France, Bibliothèque Nationale de France;

Greek 2055, fol. 53v (1401-1500)

This 11-circuit Chartres-style labyrinth is found in a

Greek philosophical miscellany, including works by

Aphrodisienis Alexander, Isidorus Hieromonachus,

Ephesisu Michael and Michael Psellus.

28 - Greek 2055, fol. 53v; courtesy of

Bibliothèque Nationale de France

Labyrinth (p.62) and manuscript:


29 - Paris, France, Bibliothèque Nationale de France;

FR 17001, fol. 2v (1401-1500)

This complex labyrinth, which resembles a modified

octagonal Chartres-style labyrinth, with ‘bastions’ at the

four corners, a cross on top and a further section below

formed of a diamond and two rectangles, is found in a

French compilation by Jean Miélot of literary and

historical texts, with translations of Cicero, Jean Boccace,

and Jean d’Udine.

29 - FR 17001, fol. 2v; courtesy of

Bibliothèque Nationale de France

Labyrinth and manuscript:


30 - Paris, France, Bibliothèque Nationale de France;

FR 17001, fol. 27r (1401-1500)

This complex labyrinth with eight segments in two large

groupings (one with five squares and one with a diamond

and two rectangles) is found in a French compilation by

Jean Miélot of literary and historical texts, with

translations of Cicero, Jean Boccace, and Jean d’Udine.

30 - FR 17001, fol. 27r; courtesy of

Bibliothèque Nationale de France

Labyrinth and manuscript:


31 - Paris, France, Bibliothèque Nationale de France;

FR 17001, fol. 88r (1401-1500)

This complex labyrinth with a pattern with nine

compartments, eight arranged around a central

square, is found in a French compilation by Jean

Miélot of literary and historical texts, with translations

of Cicero, Jean Boccace, and Jean d’Udine.

31 - FR 17001, fol. 88r; courtesy of

Bibliothèque Nationale de France

Labyrinth and manuscript:


32 - Philadelphia, PA, USA, University of Pennsylvania Library;

LJS 226, loose note 1, side 2 (1410)

This 11-circuit Chartres-style labyrinth is located in a

manuscript of astrological and astronomical texts,

originally from England and Spain.

32 - LJS 226, loose note 1, side 2;

courtesy of the University of Pennsylvania Library





33 - Paris, France, Bibliothèque Nationale de France;

Orientaux Persian 62, fol. 322v (1410)

This 9-circuit Jericho labyrinth is found in an Iranian

manuscript of Muǧmal at-tawārīḫ wa-'l-qisas, a

compendium of histories and stories.

33 - Orientaux Persian 62, fol. 322v;

courtesy of Bibliothèque Nationale de France

Labyrinth and manuscript:



34 - Dublin, Ireland, Chester Beatty Library;

Persian 322, fol. 121r (1420)

This 8-circuit Jericho labyrinth is found in an Iranian

manuscript of Muǧmal at-tawārīḫ wa-'l-qisas, a compendium

of histories and stories.

34 - Persian 322, fol. 121r;

courtesy of the Chester Beatty Library



35 - Den Haag, Netherlands, Koninklijke Bibliotheek National;

KB 72 A 23, fol. 21v (1460)

This 11-circuit modified Chartres labyrinth, with the

Minotaur at the centre, is found a manuscript copy of

Lambert of St. Omer’s medieval encyclopaedia, Liber

Floridus, from Lille in France and Ninove in Belgium.

35 - KB 72 A 23, fol. 21v; courtesy of

Koninklijke Bibliotheek National





36 - Princeton, New Jersey, USA, Garrett Library,

Princeton University; Ms. 158, fol. 157v (1471)

This 11-circuit modified Chartres labyrinth is found in

Giovanni Marcanova’s Collectio Antiquitatum on the

monuments and places of ancient Rome, probably

from Bologna, Italy.

36 - Ms. 158, fol. 157v; courtesy of

Garrett Library, Princeton University





37 - Heidelberg, Germany, Universitätsbibliothek;

Cod. Heid. Orient 118, fol. 197v (1475)

This 9-circuit Jericho labyrinth is found in an Iranian

manuscript of Muǧmal at-tawārīḫ wa-'l-qisas, a

compendium of histories and stories.

37 - Cod. Heid. Orient. 118 fol. 197v;

courtesy of Universitätsbibliothek Heidelberg





38 - Den Haag, Netherlands, Koninklijke Bibliotheek

National; KB 128 C 4, fol. 40r (1512)

This 11-circuit modified Chartres labyrinth, with the

Minotaur at the centre, is found a manuscript copy of

Lambert of St. Omer’s medieval encyclopaedia, Liber

Floridus, from Enghien, Belgium.

38 - KB 128 C 4, fol. 40r; courtesy of

Koninklijke Bibliotheek National





As libraries across the globe continue to make their manuscript collections available in

digital formats other labyrinths will come to light. We have found it useful to recheck digital

collections regularly and encourage others who are searching for manuscript labyrinths to

do the same. If you become aware of other manuscript labyrinths that are not included here,

we would be eager to learn of them, and together continue to study the development and

meanings of the labyrinth images used in medieval manuscripts.

Alain Louët, Chartres, France; April 2021. Email: [emailprotected]

Jill K. H. Geoffrion, Wayzata, MN, USA. Email: [emailprotected]

Acknowledgment & References

We would like to gratefully acknowledge the assistance of Jeff Saward in locating quality images

for this article and generally helping us to present this material in the most helpful way, and also

the many libraries and institutions that have provided images and information from their

manuscript collections online for research.

Kern, Hermann. Labyrinthe. Munich: Prestel, 1982.

Kern, Hermann. Through the Labyrinth, edited by Jeff Saward & Robert Ferré. Munich, New

York & London: Prestel, 2000. See chapter VII: Labyrinths in Manuscripts, p. 105-141.


Hedge Mazes in Portugal: A Brief History

Carlos Soreto

Gardens are amazing places where our senses enjoy the magical atmosphere created by the

aesthetics of art together with the energy of nature. The art of building gardens is very

ancient yet the inclusion of a mystical symbol like the labyrinth only began around the 14th

century in private European gardens, and in public parks a few centuries later. From the

Iberian Peninsula to Scandinavia, they flowered into a multitude of forms inspired by

medieval labyrinths and more recently by classical designs.1 As playthings of the rich and

noble, or amusem*nt places for children, the thrill they give has captivated and continues to

puzzle many generations in search of a goal that, like our inner core, proves difficult to


It seems that the taste for hedge mazes in the private gardens of aristocratic Portuguese

families begins in the 1670s, according to a notarial record from 1673

2 where there is

mentioned ‘an external walkway connecting to the oratory house which overlooks the

labyrinth’ and describes the ‘building resting on four stone columns covered by a grove of

trees and a small fountain and a pond’ at the centre of the maze that would have embellished

the gardens at the Palácio dos Marqueses de Fronteira.3 Five years later the French

intellectual Alexis Collotes de Jantillet, who lived and worked at the royal court at Lisbon,

visited the palace of Fronteira and in his book published in 1679

4 he describes the ‘beauty

and elegance of the place,’ and mentions the centre of the labyrinth as a shady spot featuring

‘four elegant columns... with water spouting from a pyramid... originating from a lake...; near

which lies a bed... where you can have a nap....’ 5 This evokes the ‘Caza Armada en quatro

Colunas de pedra Cuberta do mesmo Aruoredo’ (four stone columns shelter) mentioned in

the record of 1673, and suggests that a bower, still in vogue in the 17th

century as in the Oude

Doolhof in Amsterdam (ca. 1610),6 was a prominent feature of the baroque garden maze of

this 17th

century palace with the largest collection of Portuguese tiles in situ, located a few

kilometres from downtown Lisbon.

The maze at the Palace of Fronteira,

plan by the architect

Rodrigo Alves Dias, 1995


About 10 kilometres from the palace of Fronteira, on the road from Lisbon to Sintra, at the

royal palace of Queluz there was a fresco painting on a wall (recorded in 1772, but now lost),

depicting a game of Blind’s Man Bluff, whose diverting scenes, according to Simonetta Luz

Afonso and Angela Delaforce ‘once faced across to a labyrinth’ likely linked with the ‘idea

of hiding in the complex puzzle of the maze.’ 7

About 20 years after the Great Lisbon

earthquake of 1755,8 Richard Twiss, a British

traveller who visited the palace, wrote ‘There is

a large garden behind this palace, with a

labyrinth, and orange and lemon groves.’ 9

Additionally, several financial records (from

1767 to 1782) and the existing ground plans of

the palace kept at the Biblioteca Nacional do

Rio de Janeiro, Brazil, attest to the existence of

a hedge maze in the gardens of the royal palace

at Queluz in the 18th

century. Although the

name of the garden, Jardim do Labirinto, still

remains as a memory, the maze unfortunately

has not survived, it was probably destroyed at

the start of the 20th


Palace of Queluz, ground plan of the lower level of the gardens (detail, mid-18th century).

Garden of the Labyrinth (maze, bluish rectangle). Biblioteca Nacional do Rio de Janeiro, Brazil

Several other documents (both in paper and epigraphic inscriptions) from the 17th


onwards, mainly from nobility villas from northern Portugal, mention the word ‘labyrinth’

(labarint(h)o, labyrinto), but as with those mentioned above, all of those mazes have

disappeared over time. The only survivor, from the 18th

century, is to be found at the Quinta

da Prelada in Oporto, and in the same city there is another hedge maze at the Parque de S.

Roque, although it is of modern construction, from the second half of the 20th



Quinta da Prelada is a baroque villa located in northern Oporto, a few kilometres from the

city centre. Between 1743 and 1748 its owners, the Noronha e Menezes family,

commissioned the Italian architect Niccolo Nasoni to create a plan for building a house and

a garden including a box labyrinth (technically a maze). The maze was probably planted at

that time because it is mentioned in Diccionario Geographico in 1758, as a ‘box labyrinth

with 72 feet and three quarters in side... with a good-looking symmetry.’ 11

It was perhaps

inspired by a design in a book about agriculture by a Catalonian friar from 1617 that features

a similar square labyrinth,12

whose figure ‘may be worthwhile to those who are curious, and

inclined to similar things.’ 13

In the middle of the 18th

century João José, the first Oporto

municipal gardener, modified the gardens at Quinta da Prelada but no mention is made of

the box hedge maze.14

We have neither visitors’ accounts, as in Lisbon, nor detailed

descriptions of the role of the labyrinth in the romantic leisure atmosphere of the Quinta,

but we can surmise that according to the Christian and pious stock of the Noronha e

Menezes family, it was probably used merely as a playground for family entertainment.


A maze (without solution) that could have inspired the

maze at Prelada, in a book on agriculture by Fray

Miguel Agustín, originally published in Catalan in

1617 (reproduced here from the 1722 Spanish edition)

The maze at Prelada, plan by A. Alvão,

1917. Photo: João Baptista

The hedge maze, of rectangular shape, measures 24.5 by 32 metres and is to be found at the

rear of the house. It has three visible entrances, though initially there should have only been

one, from which the circular centre containing a 30 metres tall Araucaria tree, was reached.

We don’t know when the tree was planted, and according to a description published in 1909

by someone who was lost in the (already neglected) maze some decades previously, the

central clearing would have featured some iron benches and also a rustic table surrounding

the trunk of the tree.15

The maze is formed of four concentric circles around the central

point, leading into rectangular circuits of box shrubs ca. 1.2 metres high, with a path 90

centimetres wide and a 650 metre total length to the centre and back out. Over the years it

has undergone some alterations, and according to João Baptista its present layout does not

match the plan drawn in 1917 by A. Alvão,16

an employee of Santa Casa da Misericórdia do

Porto (the present owner of Prelada), and published in an Oporto newspaper article in 1943,

where the solution of its path drawn by a puzzle enthusiast is shown.17

The maze at Prelada, following restoration in 2013. Photo: Santa Casa da Misericórdia do Porto


After the death of the last owner of the Noronha e Menezes family, in 1903, the aesthetic

and joyful side of the Quinta was neglected as were the maze shrubs, but now, after

undergoing a restoration process before its official re-opening in 2013, it is a pleasant place

that deserves to be visited, and the maze is fully grown and once again looking splendid,

waiting to be threaded... all for free.

S. Roque

Not too far from Quinta da Prelada there was another villa, that once belonged to the

wealthy Port wine traders Ramos Pinto and Calem family, the Parque de S. Roque. The

palace-like house built in 1792, and the rest of the park, was acquired in 1978 and 1979 by

the Oporto municipality and on 20th

July 1979 the park was officially opened to the public.

Arranged in a series of stepped terraces with an area of 45,000 square metres, it is located

on a steep hill facing the Douro river, near the train station of Campanhã.

The circular box hedge maze of S. Roque. Photo: João Baptista.

The luxuriant box hedge maze, that according to some testimonies would have been planted

around 1985,18

is to be found near the main park entrance, on Rua de São Roque da

Lameira, on a formal garden platform that provides its rectangular shape, bordered by

camellias. The maze is of concentric type with a granite pillar surmounted by a teardrop

shape ornament at its centre and measures ca. 30 metres in diameter with well-tended

shrubs ca. 60 centimetres wide and 1.5 metres high. The path, approximately one metre

wide, can be entered at two different points. The park of S. Roque is open daily and can be

visited in winter from 8 a.m. to 7 p.m. and in summer from 8 a.m. to 8 p.m., with the free


Carlos Soreto, Tocha, Portugal; April 2021

Email: [emailprotected]




I would like to take this opportunity to thank João Baptista, architect and labyrinth enthusiast

whose correspondence exchange has awakened in me an old wish that finally came true.

Moreover, such an elaborate article would not have been possible without his invaluable help.

My thanks also to Eng. Paula Aleixo of Santa Casa da Misericórdia do Porto who kindly put at

my disposal the photograph of Casa da Prelada hedge maze.


Afonso, Simonetta Luz, and Angela Delaforce. Palace of Queluz: The Gardens. Lisbon: Quetzal;

IPPC, 1989.

Agustín, Fray Miguel. Libro de los secretos de agricultura, casa de campo, y pastoril; Traducido de

Lengua Catalana en Castellano, por Fray Miguel Agustin... del Orden, y Religion de San Juan

de Jerusalén, del Libro, que el mismo Autor sacó à luz el año de 1717 [i.e., 1617]. Y ahora con

addicion del quinto Libro... y un Vocabulario de seys Lenguas... con una Rueda perpetua, para

conocer los años abundantes, ò estériles. Barcelona, 1722, p. 52.

Andresen, Teresa, and Teresa Portela Marques. Jardins Históricos do Porto. Lisbon: Edições

Inapa, 2001.

Araújo, Ilídio de. Arte paisagista e arte dos jardins em Portugal. Lisbon: Direcção Geral dos

Serviços de Urbanização, Centro de Estudos de Urbanismo, 1962. Vol. I.

Baptista, João David Esteves. As meditações de Dédalo: labirintos e arquitectura. Coimbra, 2005.

Carapinha, Aurora da Conceição Parreira. Da essência do jardim português. Évora: Universidade

de Évora, 1995.

Carita, Helder, and António Homem Cardoso. Tratado da grandeza dos jardins em Portugal ou da

originalidade e desaires desta arte. Lisbon: Círculo de Leitores, 1990.

Dias, Rodrigo Alves. “O Giardino Segreto: percorrendo um jardim virtual.” Monumentos 7 (Set.

1997), p. 31-35.

Matthews, W.H. Mazes and Labyrinths. New York: Dover, 1970, p. 150.

Saward, Jeff. Labyrinths and Mazes. London & New York: Gaia/Lark Books, 2003, p. 153-177.

Soreto, Carlos. “Portuguese Mosaic Labyrinths.” Caerdroia 33 (2003), p. 33-39.

Twiss, Richard. Travels Through Portugal and Spain in 1772 and 1773. London, 1775, p. 22.

Viterbo, Sousa. Diccionário histórico e documental dos architectos, engenheiros e constructores

portuguezes ou a serviço de Portugal. Vol. II. Lisbon, 1904, p. 192 & 193.


1. For instance, the design of the hedge maze planted in 1981 at Cawdor Castle, Scotland, is

inspired by the most famous of the four labyrinths found on the Roman mosaics at

Conimbriga, Portugal [Soreto 2003].

2. Transcribed in Carita & Cardoso, 1990, p. 107-108.

3. ‘... uma baranda de pasejo descuberta que fica sobre o Labarinto que uaj entestar na caza

do oratorio...’. (Tombo de 1673, fol. 93).

‘... e no mejo do labarinto de [árvores de] espinho ha huma Caza Armada en quatro Colunas

de pedra Cuberta do mesmo Aruoredo e huã fonte pequena Com seu tanque tudo

embrexado...’. (Tombo de 1673, fol. 95).


4. Extract transcribed in Carita & Cardoso, 1990, p. 110.

5. ‘... Um lugar umbroso tecido de frondosos e sempre verdes raminhos e abobadado sôbre

quatro agradáveis colunas de elegante fabrico; uma priâmide feita de fragmentos de vidro e

de louça chinesa, espécie de ornato coberto de conchas, conchinas e pérolas entre outras

cousas, expele água em forma de coroa aberta, dando origem a um lago que é ornamentado

de várias côres; perto está um leito preparado onde te deitarás se te apetecer dormir a sesta

em dias de Verão, conciliando o sono com o som da água caindo lentamente’. (Extract of

Alexii Collotis de Jantillet Horæ subsecivæ, Ulyssipone: ex typographia Joannis a Costa,

1697, originally translated and published by José Cassiano Neves in his book Jardins e Palácio

dos Marqueses de Fronteira, in 1941).

6. Saward, 2003, p. 159.

7. Afonso & Delaforce, 1989, p. 33.

8. This devastating earthquake occurred in 1755 was known everywhere in Europe and as a

reminiscence of the confusion at that time, in the neighbourhood of Viborg, Finland, one of

the names by which stone labyrinths were known was ‘Lissabon’ [Matthews 1970].

9. Twiss, 1775, p.22.

10. Afonso & Delaforce, 1989, p. 33.

11. ‘Pegado ao segundo Jardim ha um Labarintho de buxo, que tem setenta e dois pez e tres

quartos por lado, a arte lhe fez com alguas figuras do mesmo buxo, mais vistosa a symmetria’.

(Diccionario Geographico, tom. 31, fol. 25, transcribed in Viterbo [1904, 192]).

In the same folio of the manuscript (transcribed in Viterbo [1904, 193]), is mentioned

another villa (Quinta do Viso) not too far from Prelada, with ‘a good garden and labyrinth’

now disapeared, also designed by Niccolo Nasoni, the architect who planned the Oporto’s

landmark Torre dos Clérigos, among many other secular and ecclesiastic emblematic

buildings of baroque architecture in northern Portugal. According to Sousa Viterbo (A

jardinagem em Portugal: Apontamentos para a sua história. Coimbra: Imprensa da

Universidade de Coimbra, 1906, p. 48) this villa was a replica of Quinta da Prelada.

12. The design is reproduced side by side with Prelada’s in Carapinha, 1995, Vol. II, p.76.

13. ‘... pondremos aqui una figura, de la qual traza se pueden valer los que son curiosos, è

inclinados à cosas semejantes’. Fray Miguel Agustín, Libro de los secretos de la agricultura,

casa de campo y pastoral, 1617.

14. Carita & Cardoso, 1990, p. 296.

15. ‘A meio do labyrintho havia uma clareira, ao centro da qual se erguia uma arvore rica de

boa sombra, cujo tronco era rodeado por uma meza rustica, em annel’.

‘Alguns bancos de ferro em redor, pareciam convidar-nos [...] ao descanso de tão penosa

aventura’. (O Tripeiro 1909, Série 1, Ano 2, Nº 37, p. 15).

16. Personal communication.

17. O Primeiro de Janeiro, Nº 227, 1943, p. 1.

18. Campanhã: estudos monográficos. Porto: Junta de Freguesia de Campanhã, Câmara

Municipal, 1991.


The Surroundings of the Rösaring Labyrinth

Heather Robertson

Rösaring is one of Sweden’s most impressive stone labyrinths and probably the oldest.

Although quite near Stockholm, its surroundings are unusually intact and may have much

to tell us about the earliest uses and dating of stone labyrinths. Börje Sandén (1928 - 2020)

was official guide for many years and introduced thousands of people to this labyrinth. As

nearby discoveries unfolded over the decades, he also introduced visitors to a remarkable

cult site adjacent to the labyrinth, and to an ancient settlement immediately below. His

enthusiasm was catching – only he could inspire a hundred visitors to crowd around and

stare at a flat piece of green grass just because it was once the site of a bronze-caster’s

workshop. There was so much to see there, he would say, that no guided tour could cover it

all – and none of his tours was like any other. His main paper1 and other writings

2 convey a

detailed knowledge of the unique surroundings of Rösaring, along with his deliberations

over rites that once took place there.

Fig. 1: In 2008 the Nordic Museum awarded Börje and

Gudrun Sandén the Hazelius Medal for many years of

local heritage work, including setting up a research

institute in 1987. Photo: Kjell Nilsson/Ledungen

Börje Sandén first visited Rösaring in August 1952,

a week after he and his wife Gudrun (figure 1)

moved to the district to begin work as teachers.

Back then the cult road close to the labyrinth was

still undiscovered, hidden by extremely slow-

growing pine forest. The labyrinth itself was only

just discernible, but later it was cleared of

vegetation and a fence erected to protect it. In 1968 Börje was appointed local

representative by the Swedish National Heritage Board. His long association with Rösaring

provided continuity as many different researchers, amateurs, officials, and sign-writers came

and went. He developed three original ideas which may eventually lead to a better

understanding of this enigmatic and only partly investigated archaeological site.

1) The cult site was better suited to fertility rites, such as the “goddess in wagon”

ceremonies described by the Roman writer Tacitus about 98 CE, than to Viking Age

burials as first thought. His interpretation agrees with legends associated with Swedish

labyrinths and with the labyrinth being an integral part of the cult site.

2) There may be a connection to a large 4000-year-old hilltop cult site to the north,

suggesting two stages of construction at Rösaring and an even earlier date for the


3) Interaction was likely between the settlement below Rösaring and the nearby Viking

Age proto-town of Birka, where Christianity was first introduced to Sweden.


The Labyrinth, Cult Site, and Settlement at Sanda

The Rösaring labyrinth (figure 2) differs from most of Sweden’s 300 stone labyrinths in

several ways. It has 16 circular walls, rather than 12 or 8, and is not situated on a beach or

near a church. It has the earliest documentation3 and may belong to a special group of about

21 labyrinths that once were focal points for communities in the oldest agricultural districts

of Sweden.4 Most unusual of all is the cult site (figure 3) found in its surroundings, recently

named “The Rösaring Complex.”

Fig. 2: The



Drawing by

John Kraft,


The cult site has unexcavated mounds and cairns, and a

540 metre long, 3 metre wide, ridge-top road, edged on

both sides with stones and topped with clay. The road

was discovered in 19795 and was thought to have been

used for funeral processions.6 A large mound close to

the labyrinth marks one end of the road, which runs

north along the ridge and ends at the base of a small

building. The road is flanked on its west by a ditch that

provided gravel for building it up level. Along its east

side are approximately 100 shallow round depressions 1

metre wide and 4 to 5 metres apart, for which no

explanation was found. The only carbon sample came

from the south end, a few metres from the base of the

large mound and was dated to 9th

century CE. About 400

metres from the labyrinth, on fertile land below the

ridge, was a settlement at Sanda that lasted from the

Late Bronze Age to Viking times.7 It covered an area of

at least 30,000 m² and included a bronze-caster’s

workshop, uncommon in Sweden and indicative of high

status. Today the land is used for farming. Both the cult

site and settlement may hold clues to the original

purpose of the labyrinth, as well as its age.

Fig. 3: The Rösaring Complex. Drawing from Löthman and Winberg, 1981


A Cult Site for Burials or Fertility Rites?

Whether the Rösaring Complex was designed for funeral processions may become clearer

if the large flat-topped earth mound near the labyrinth is excavated. Funerals were

suggested by graves to the south, and by the southern end of the cult road seeming to

“disappear” into the mound (figure 4), as if a departed chieftain had been conveyed in state

along the road to his final resting place. The building in the

north has been interpreted as a mortuary where funeral

processions began. A Viking Age burial in the mound and a

related role for the labyrinth, such as assisting transition to the

afterlife, would suggest an age of up to 1200 years. Signboards

at Rösaring have long featured funeral processions, some

taking the interpretation to unlikely lengths, showing the cult

road lined by fires, posts, and statues, none of which is

supported by archaeological findings. No charcoal was found

in the round depressions, and the soil under them was

undisturbed at a depth of 100 to 200 mm. This means no posts

or other tall objects ever stood in them.

Fig. 4: Excavation of the south end of the cult road,

flat-topped earth mound at left. Photo: Börje Sandén

Börje Sandén’s own interpretation of the site is no less colourful but takes account of local

geology as well as archaeological findings. He questioned the burial explanation, pointing

out that the clay-silt (mjäla) topping the mound is known to slump readily, due to a very fine

particle size. With centuries of rain, the finest material would drift down and spread out,

giving an impression of the road entering the mound. He noted that the mound resembled

another at Old Sigtuna,8 that was found not to contain a burial, but rather it was a thing

(ting) mound, with a flat top for speakers at open air courts. Further, he pointed out that

the carbon sample might not indicate the road’s construction date, but the last time it was

surfaced. Abandoning both burial and Viking Age themes, Sandén based his interpretation

of the Rösaring Complex on descriptions of Germania by Tacitus, writing about 98 CE.9

Customs of the powerful, seafaring Svione, the Svear people from whom Sweden takes its

name, were said to include the earth goddess Nerthus travelling about in a covered wagon

drawn by cows to bring peace and a good harvest. This interpretation agrees with the long-

standing association of Scandinavian labyrinths with goddesses and fertility rites.10


ceremony ended with ritual cleansing of the wagon, covering and “the goddess herself,”

carried out by slaves at a “secret lake” in which they were later drowned. A possible

connection with such a ceremony is strengthened by the cult road pointing north to a

geological feature that would have served such a purpose. The unusually steep sided valley

of Djupdal is just over a kilometre away. Due to land elevation, it is now a valley on the

Mälar Lake shore, but 2000 years ago when Tacitus was writing it was a semi-enclosed bay

of the Baltic Sea, which to observers on land would appear as a lake. A fertility ceremony

might have started at the cult road’s northern end, with the rectangular enclosure (4.8 x 6.4

metres) serving as a wagon shed, rather than a mortuary. A procession would then move

south towards the labyrinth – perhaps the focal point of the main ceremony – afterwards,

returning north to the secluded bay for cleansing, then back to the shed ready for next year.


Sandén noted that unclear points in Tacitus’ text made such a place hard to locate, but his

proposal for the Rösaring Complex agrees well with its layout. It also agrees with deities

indicated in place names typically found near older stone labyrinths, such as the earth

goddess Härn or Njärd (called Nerthus by Tacitus) and the sky god Ull, and with the idea of

their union in spring bringing fertility to the land. “Vi” means a sacred place, and the names

Härnevi and Ullevi are found just a few kilometres from the Rösaring labyrinth. Near

Härnevi and Stora Ullevi are rock outcrops with cup marks,11

while amulet rings

(figure 5) were found near Lilla Ullevi.12

The labyrinth was still used by young people in

1717 for dancing in summer,13

more likely an echo of old agricultural customs than of

burials. If fertility rites with wagons at the time

of Tacitus were once the purpose of the

Rösaring Complex, the labyrinth would be an

integral part of the cult site constructed together

with the road, with an age of just over 2000 years.

This would mean that nearby Sanda had already

been settled for several hundred years when the

cult site was built.

Fig. 5: Some of 65 amulet rings found at Lilla Ullevi.

Photo: Matthias Bäck/RAÄ

Oldest Possible Date for the Labyrinth

Cairns near the labyrinth may show the cult site was in use even earlier if they contain

Bronze Age burials, which were typically made on high rocky places, not farmland. An older

date may also be indicated by a slight change of direction in the cult road one third of the

way along its length. This was disregarded by others,14

but Sandén noted it could mean the

road was built in two stages. Its northern part points to the “secret lake,” but its southern

part points to the 4000-year-old cult site of “Draget,” at Ullfjärden (figure 6). Assuming two

construction stages, the labyrinth is closest to a likely first stage near the cairns. This part of

the road might have been oriented to connect with the larger site by a line of fires in

fireproof dishes in the round depressions, pointing to two sacred fires either side of Draget’s

entry portal. The fine detail is yet to be


but if the line passes exactly between

them, Sandén’s theory would gain ground and

raise the possibility that the labyrinth was

associated with earlier rites than the

ceremonies described by Tacitus. A new cult

may have developed from the old one.

Assuming the labyrinth was built together with

the roadway, it could be almost as old as the cult

site of Draget, quite possibly over 3000 years

old. If so, the Rösaring cult site may have been

established at the same time as Sanda or even


Fig. 6: Map of Rösaring and near

surroundings. Drawing by H. Robertson


Strategic position of the Rösaring labyrinth

For thousands of years the Rösaring Complex overlooked a major sailing route in the Baltic

Sea that passed many important places (figure 7). In the Bronze Age, open water led past


the largest hilltop cult site in Uppland. By 800 CE the route included the Viking

Age proto-town of Birka, once a powerful trading centre and nowadays an important

Swedish archaeological site. Sandén noted that Birka would have been visible from Rösaring

and that there must have been interaction between the two settlements. Each had a bronze-

caster’s workshop, probably established earlier, and for longer at Sanda, and only two other

places in Sweden, Sigtuna and Lund, had such workshops. It was in Birka’s early years that

the cult road may have been surfaced, perhaps prompted by increased travel and knowledge

of roads in other lands. Christianity was introduced to Sweden at Birka, while there are

several indications that people

living near Rösaring kept to the

old pagan ways longer than


Sanda continued for a

few hundred years after Birka’s

role was taken over by Old

Sigtuna. Around 1200 CE the sea

route began to develop strong

currents and former harbours

disappeared due to land

elevation. Rösaring became

increasingly remote from both

sea and road traffic, which may

account for its good state of

preservation today.

Fig. 7: Rösaring on the old sea

route. Drawing by H. Robertson

A Legacy of Ideas

By pulling together many apparently unlikely threads – combining old writings, local

geology, and modern archaeology – Börje Sandén has left us with new ideas for research

that may help shed more light on the stone labyrinths found in the earliest agricultural

districts of Sweden. The Rösaring labyrinth appears to have been given pride of place on

the highest part of a major glacial ridge, suggesting that it was intended as an important

element of the cult site right from the start. We still do not know for sure when that start

took place, but the surroundings point to a time well before the Viking Age. Further study

of mounds and cairns, the cult road, old settlements and shorelines, and the cult site Draget

may complete the picture.

Heather Robertson, Wensleydale, Victoria, Australia; May 2021

Email: [emailprotected]




1. Börje, Sandén. “Fifty Years with the Cult Site of Rösaring.” Viking Heritage Magazine,

3/2002. Gotland University, online at:

2. For example, see – Börje Sandén. Excursions to Rösaring ridge in 2019 (6

pp.), The Rösaring Complex. Guided tour 19 May 2019 (16 pp.), and Arkeologidagen 2009 –

Lilla Ullevi (5 pp.), also excerpts from his book Det Hände i Upplands-Bro.

3. John Kraft. “Built in Honour of Odin and Danced Around.” Caerdroia 46 (2017), p. 9-16.

4. John Kraft. “Trojeborgar, kultplatser och städer.” Östergötland, Meddelanden från

Östergötlands och Linköpings Stads Museum, Linköping. 1980, p. 75-105.

5. In 1979, a small boy walking on the ridge spotted a line of stones and his grandfather, a

farmer, then saw a parallel ditch and asked why it was needed on such a well-drained, porous

site on a glacial ester. An official report followed. Börje Sandén, as told to Heather

Robertson in 1998.

6. David Damell. “Rösaring and a Viking Age Cult Road.” Archaeology and Environment 4, in

Honorem Evert Baudou,1985, p. 171-185.

7. Peter Bratt and Kjell Andersson. Arkeologiska undersökningar vid Rösaring, Sanda och Stora

Ekeby. Låssa Socken, Upplands-Bro kommun, Uppland. Rapport 2000:11, Stockholms Läns

Museum, 2000.

8. Old Sigtuna was ca. 4 kilometres west of present day Sigtuna, at the manor of Signhildeberg

in Upplands-Bro, as revealed in documents found by Sandén when researching folk music.

This led to five summers of archaeological excavations and a revised history of Sigtuna.

9. Tacitus. The Agricola and the Germania. Chapter 40. Penguin Classics, 1970. Relevant

passage quoted in Sandén’s main paper, see note 1.

10. For example, John Kraft. The Goddess in the Labyrinth. Religionsvetenskapliga skrifter

nr.11. Åbo: Åbo Akademi, 1985.

11. Cup marks up to 10 cm wide and 2 cm deep, cut into rock outcrops, are typically Bronze

Age, and occur in the garden of a house near Härnevi. Such marks are said to have been

greased and used for metal offerings by “wise old women” into modern times. In 1981 a coin

from 1718 was found in a cup mark near Stora Ullevi, under lichen and moss. See note 2.

12. Ullevi occurs twice here, combined with Stora (large) and Lilla (small). At Lilla Ullevi in

Bro township, no cup marks were found but excavations in 2007 yielded 65 iron amulet rings

dated to the Vendel period (ca. 540-790 CE). These appeared to have been hung on poles

in a specially marked area, possibly linked with swearing of allegiance to Ull.

13. Johannes Arenius. Fjärdhundra, Dissertation, Uppsala, 1717, p. 68.

14. Emília Pásztor, Curt Roslund, Britt-Mari Näsström and Heather Robertson. “The Sun and

the Rösaring Ceremonial Road.” European Journal of Archaeology, Sage Journals, Vol. 3

Issue 1, 2000, p. 57-61.

15. Sandén had planned to check the alignment in detail using GPS during 2020.

16. “Draget” is where boats were dragged across land into a seaway north of Rösaring called

“Ullfjärden,” when the route was no longer fully covered by water, around 400 or 500 BCE.

17. At Härnevi, cup marks were found at 15 metres elevation, but at 20-25 metres elsewhere in

Uppland. By the late 1100s, the Christian parish had been given the neutral name of “Bro”

rather than the local village name of Härnevi, perhaps because a competing pagan cult was

still in existence. See note 2.


Historic Turf Maze Sites in Wales

Jonathan Mullard

In the introduction to his article on the Llwydiarth Hall Labyrinth in Caerdroia 49, Jeff

Saward mentions that, until its discovery in 1995, there were no locations known for turf

mazes in Wales. This despite numerous early references to the topic in general. Perhaps

many of these mazes, “sometimes cut out in the turf by shepherd boys whilst they are tending

their flocks on the mountains of Wales,” were just too ephemeral to be recorded [Lowe 1924].

Leaving aside the question, which has never been properly addressed, as to whether

shepherd boys really had enough time, the tools and motivation, to cut relatively complex

patterns in the turf, it is possible, as has been done elsewhere, to identify likely sites in Wales

though place names.

Surprisingly little work has been done on this aspect to date, but one of the great benefits of

the internet is the easy availability of information that could only previously be extracted

through the slow exploration of libraries and archives. Today the researcher interested in

Wales, and Welsh affairs, is well served by several websites, not least the three excellent sites

curated by the National Library of Wales: Welsh Journals, Welsh Newspapers and Welsh

Tithe Maps. Welsh Journals, for instance, provides access to journals relating to Wales

published between 1735 and 2007. Titles range from academic and scientific publications to

literary and popular magazines and it is possible to instantly search over 450 journals, in

both English and Welsh; a total of some 1.2 million pages.

While I have been using these websites for a number of years, in connection with my natural

history interests, it is only recently that I have started searching them for references to turf

mazes in Wales. Exploring the Welsh Newspapers website, for instance, I found an

interesting item in the ‘Archaeological Notes and Queries’ section of the issue of The Weekly

News and Visitors' Chronicle for Colwyn Bay, Colwyn, Llandrillo, Conway, Deganwy and

Neighbourhood for 17th February 1899. Here the Rev. Meredith J. Hughes, F.R. Hist.S, is

replying to a previous enquiry from ‘G.P.J’ about turf mazes.

“The ancient Welsh regarded Brutus, a Prince of Troy, as the original founder of their

nation. It is curious to note that they sought to propagate their tradition by cutting a

plan of Ilium, the chief city of Troy, in the greensward and on mountain sides. The

Welsh for Ilium is Caerdroia - The City of Troy, and it is conjectured that the suffix-

droia, which in Welsh means ‘to circle’, suggested to them the interesting plan showing

in the accompanying diagram. These plans of Ilium were very common in the

Principality [that is Wales] some few centuries ago. The second diagram shows the

remains of one of these mazes and may be seen at Mynydd Merci, about a mile to the

S.W. of the new Church now being built by Mrs. Frost, at Bryn-y-maen. I have

previously traced here the lines of an extensive British hill- camp, and my conjecture is

that some of these lines were made use of in delineating the geometrical design of Ilium.

It is to be regretted that some of the lines cannot be accurately traced owing to various

causes, chiefly the fact that much of the land is under cultivation.”


Fig. 1: The ‘interesting plan’ referred to by the Rev. Meredith J.

Hughes together with his sketch showing ‘the remains of one of

these [turf] mazes’

The Reverend’s assumption of a ‘British hill-camp’, what

we would now call a hillfort, appears to be mistaken but

the sketch of a possible turf maze (figure 1) is the only one

found so far in Wales (OS Grid Ref: SH 82523 76108).

Intriguingly, Reverend Hughes goes onto say that:

“Mr. Venables Kirk writing to the Arch. Cam.

mentions a similar labyrinth not far from his beautiful

home - Nantyffrith, near Wrexham. That instance,

however, is much more elaborate and complete, than

the one under notice.”

‘Arch. Cam.’ is Archaeologia Cambrensis, the long-

running Welsh antiquarian or archaeological journal and

‘Mr Venables Kirk’ is actually Richard Venables Kyrke of

Nant y Ffrith, Wrexham. Nant y Ffrith Hall was originally

built as a hunting lodge in 1850, with successive owners

enlarging the building and creating landscaped gardens. It

eventually became derelict and was demolished between

1947 and 1950 (OS Grid Ref: SJ26515418). Kyrke,

although “an authority on the Roman roads and British

Camps of his district” apparently published nothing except

a single letter in Archaeologia Cambrensis. There are in

fact two letters in the journal from Kyrke, but neither refer to a labyrinth (Kyrke 1871/1879).

In his second letter of 27 December 1878 though he mentions that near the Hall “is a place

which puzzles me very much.” Referring to “a number of parallel trenches” he states that they

are “on comparatively level ground, and wind round in a kind of an arc of a circle, following

the hill above.” Since they were “3 or 4 feet deep” and the largest “would be wide enough for

a cart to go along” they were certainly not the remains of a maze. This maze was possibly the

result of the Reverend Hughes’ imagination, unless of course there is another record which

has not yet come to light.

One of the documents located on the Welsh Journals website is a paper by John Hobson

Matthews on The Placenames of the Cardiff District, which was read before the

Archaeological Section of the Cardiff Naturalists Society and published in their Report and

Transactions for 1900-1901. In this paper Matthews notes that:

“Many of us remember the tumble-down cluster of old cottages which stood on the site

of this Free Library and Museum. Those small houses, and the old Calvinistic

Methodist Chapel, were built on a piece of ground which was known as ‘Little Troy’. A

volume might be written on the English placename ‘Troy’; but it will be sufficient to say

that it meant a maze, and that mazes were commonly found in the precincts of parish

churches. They had a symbolic significance. Sometimes a pictorial maze was

represented in painting or mosaic upon the walls of the church. It is possible that the

maze was identical with the whole or part of the Trinity garden already referred to.”


“Trinity Street received its name from the Guild of the Holy Trinity, which, previous to

the Reformation, was attached to Saint John’s Church, and held a good deal of house

and land property close by – among them the Trinity Garden, the site of which is now

covered by the flagged space in the Hayes and by the building in which we are now

assembled [the Free Library and Museum].”

Furthermore, Matthews goes onto say; “‘Hays’ I need hardly remark, means an open tract of

grassy land. The Hayes… was a place of that sort down to a hundred years ago.”’ So, we know

that before 1800 there was a grassy area right in the centre of Cardiff that had the name

‘Little Troy’ and was owned at one time by the Guild of the Holy Trinity. Today the area is

still ‘flagged’ and pedestrianised;

and known for its open-air snack

bar. Although it now seems an

unlikely location, at one stage in the

area’s history there was probably a

turf maze here (OS Grid Ref:

ST18374 76484).

Fig. 2: An extract from an 1880 map of

Cardiff showing St John’s Church and

the old Calvinistic Methodist Chapel,

together with a number of small houses,

on the site of a possible turf maze

In the Middle Ages, Cardiff, like other towns, possessed Guilds of various kinds, of which

the most important were those of the merchants and traders. King Edward II, on the same

day that he gave a Charter to Cardiff, on 4 March 1323/4, granted rights and privileges “to

the burgesses of the arts or crafts of Cordwainers and Glovers of the town of Cardiff and to their

successors for ever.” ‘Cordwainers’ were shoemakers and are noted as one of the most

significant Guilds in Cardiff at this date. The various Shoemakers Guilds have an intimate

connection with a number of turf mazes, not least the Shoemakers’ Race that once existed

on Kingsland in Shrewsbury – and which I uncovered as part of my researches nearly 40

years ago [Mullard 1983]. Similarly, there are numerous mentions of the Shoemakers’ Hall,

which was located on the, once important, thoroughfare known as Shoemaker Street, which

formerly came out into Saint John's Square, near to the site of the possible maze.

John Speed’s Plan of Cardiff (1610) depicts individual buildings including the castle, town

hall, churches and former priories, together with the old walls and gateways, and names

streets but there is no indication of a maze on what became known as The Hayes. Neither

do other, later, maps include any reference to a maze, but large-scale maps of towns were

commonly focused on buildings, so a turf maze may not have been thought worthy of

recording. The church dedicated to St John the Baptist is the oldest building in Cardiff still

in continuous use, apart from the nearby castle. Like many churches it was heavily restored

in the Late-Victorian Period, but, at heart, it remains essentially a medieval building. An

initial exploration a few years ago unfortunately did not reveal “a pictorial maze… in painting

or mosaic upon the walls of the church,” but there might possibly be a representation

somewhere in the Cardiff archives.


Fig. 3: A photograph of the possible maze site in the

centre of Cardiff in 2013, showing what is now the Old

Library, which houses the tourist information and

exhibition centre, and the snack bar. To the left is the

parish church of Saint John the Baptist. Copyright

Robin Drayton and reused under the Creative

Commons Licence

There are doubtless other possible Troy

references, and turf maze sites, to be found by

searching the Welsh language journals;

although many of these references relate to the

legend of Troy and links with Wales.

As an example of what can be found elsewhere, a search of the Welsh Tithe Map database

reveals, on an 1841 map, Caer-droia Cottage and Garden, tenanted by Ebenezer Evans, in

Llanfair Orllwyn, near Horeb. Tantalisingly, the large garden (Field 476, covering 2 Roods

and 30 Perches – approximately 0.7 acres) is recorded as ‘pasture.’ Was there a turf maze

cut into this pasture? Was it cut by shepherd boys tending their flocks? We may never know.

Sadly, the Cottage has long disappeared, along with the nearby field boundaries, but the

shape of other boundaries can still be traced on the Ordnance Survey map, which enables

the location to be confirmed (OS Grid

Ref: SN38277 42661). It is surprising that

only one such reference has been found

to date but it indicates perhaps that the

name was only used in relation to turf


Fig. 4: Extract from the Tithe Map for

Llanfair Orllwyn in the County of Cardigan,

now Ceredigion, showing Caer-droia Cottage

and Garden

Yet more potential maze sites can be found by searching The Royal Commission on the

Ancient and Historical Monuments of Wales (RCAHMW) List of Historic Place Names of

Wales; which is the first statutory catalogue of such names in the world. The Historic

Environment Wales Act 2016 requiring Welsh Ministers to compile and update the list on

a regular basis. The database is still being added to but a search for ‘maze’ reveals, on the

Ordnance Survey Second Edition six-inch

to the mile map, not a turf maze but a

former hedge maze at Highmead House

in Llanwenog, in the old county of

Cardiganshire. (OS Grid Ref: SN 49995


Fig. 5: Extract from Ordnance Survey Second

Edition six-inch to the mile map showing a now

destroyed hedge maze at Highmead House


A similar hedge maze was once associated with Allt-y-ferin in Llanegwad. A mansion built

around 1860, it was demolished in the 1950s (OS Grid Ref: SN 51939 22771). Finally, there

was a maze site in the grounds of Gwernyfed Park in Tregoyd and Felindre (OS Grid Ref:

SO 17798 36783). This last site, in what used to be Garden Wood but is now just open

grassland, is not pictured on the map like the other two, so could just possibly have been a

turf maze. Here at Old Gwernyfed are preserved the unusually extensive earthworks and

architectural remains of an Elizabethan-Jacobean terraced, formal garden so it is likely that

any maze was associated with these. Unfortunately, from aerial photographs it is clear that

none of the mazes associated with these large houses survive today.

A more definite turf maze site is suggested by a field called the Race Piece in Kerry, in the

former county of Montgomeryshire, since this is a name associated with turf mazes just over

the border in Shropshire (OS Grid Ref: SO 14405 88402). ‘Troy’ is not a common field name

in Wales, so Cae Troy, a field in Glascwm, in the former county of Radnorshire, recorded

in 1841 may well indicate another site (OS Grid Ref: SO 13027 53978). Similarly, Troy Piece,

a field in Llananno, Radnorshire is also a likely candidate for a turf maze (OS Grid Ref: SO

08235 76479). I have ignored the seemingly obvious candidate in Wales for a turf maze, the

village of Mitchel Troy in Monmouthshire, since there is good evidence that the English

name derives from the name of the river which passes through the settlement, the Welsh

‘Troddi,’ becoming ‘Trothy’ and then ‘Troy.’

Although it is not a turf maze, to complete this initial survey of Welsh labyrinths it is worth

mentioning the curious ‘maze’ at Gwydir Uchaf on the edge of Gwydir Forest in Snowdonia

(SH 79507 60975). Constructed in the

1600s, it consists of a spiral hedge

running around a mound and is part of

the garden associated with the nearby

Gwydir Castle. The feature used to be

overgrown, but was restored in 2003. The

nearby chapel is a Cadw site.

Fig. 6: Extract from Ordnance Survey Second

Edition six-inch to the mile map showing the

spiral hedge feature next to the Gwydir Chapel.

Further research will obviously be necessary on all the turf maze sites mentioned and indeed

the hedge mazes. Since initial explorations can be done from the comfort of your home, I

would be interested to hear of further discoveries readers may make in Wales using the

websites I have described. As well as these historic turf maze sites there are a number of

modern examples in Wales and it is proposed to cover these in a later article.

Jonathan Mullard, Hexham, England; May 2021

Email: [emailprotected]

Jonathan Mullard is the author of a small booklet, Caerdroia Salopia, published in 1983,

which first revealed the existence of former turf mazes in Shropshire. He is now updating

this publication, while also researching a larger work on similar features in Wales.


Provisional List of Historic Mazes and Labyrinths in Wales

Type Name Location OS Grid Reference

Turf Merci Maze Mynydd Merci, Conwy SH 82523 76108

Turf Little Troy City of Cardiff ST18374 76484

Turf Caer-droia Llanfair Orllwyn, Ceredigion SN38277 42661

Turf Race Piece Kerry, Powys SO 14405 88402

Turf Cae Troy Glascwm, Powys SO 13027 53978

Turf Troy Piece Llananno, Powys SO 08235 76479

Hedge? Llwydiarth Labyrinth Llanfihangel, Powys SJ 05980 16630

Hedge Highmead Maze Llanwenog, Ceredigion SN 49995 43162

Hedge Allt-y-ferin Maze Llanegwad, Carmarthenshire SN 51939 22771

Hedge? Gwernyfed Maze Tregoyd and Felindre, Powys SO 17798 36783

Hedge Gwydir Uchaf Maze Llanrwst, Conwy SH 79507 60975

NB: Grid References are taken from the Ordnance Survey online map. Coordinates for turf

mazes are taken from the approximate centre of the relevant, historic, field.


Hughes, M.J. “Query - Archaeological Notes and Queries” The Weekly News and Visitors'

Chronicle for Colwyn Bay, Colwyn, Llandrillo, Conway, Deganwy and Neighbourhood, 17th

February, 1899.

Kyrke, R.V. “Reports Roman remains from Ffrith.” Archaeologia Cambrensis, 1871, p. 97-98.

Kyrke, R.V. “Letter on trenches near Nantyffrith.” Archaeologia Cambrensis, 1879, p. 155.

Lowe, W.B. “Some arrow stones and other incised stones in North Carnarvonshire and North

Denbighshire” Archeologica Cambrensis, 7th Series Vol. 4, 1924, p. 393.

Matthews, J.H. “The Placenames of the Cardiff District.” Cardiff Naturalists Society Report and

Transactions for 1900-1901, p. 33-36.

Matthews, J.H. (ed.). “Records of the Cordwainers and Glovers: Introduction” in Cardiff Records,

Vol. 3, 1901, p. 336-341.

Mullard, J. Caerdroia Salopia - The Lost Turf Mazes of Shropshire. Mizmaze Publications, 1983.

The spiral maze at Gwydir Uchaf was originally

planted in the early 1600s and restored in 2003.

Thanks go to Stephen Shaw for spotting this one.

Photo from Google Earth, 2015


From Jerusalem to Troyborg: The Labyrinth Name Change of the North

Christina fa*gerström

The names of the stone labyrinth monuments found in northern Europe could help us to

understand their use, typology, wide area of distribution and timeframe. The labyrinth names I

will focus on in this paper are Jerusalem and variations on the theme of Troyborg. In Scandinavia

these names are understood to refer to the same typology used for outdoor stone monuments;

but a Jerusalem and a Troyborg would seem to refer to widely different cultural ideas and


As the heading ‘From Jerusalem to Troyborg’ implies, I am suggesting that the name Jerusalem

for a stone or turf labyrinth precedes the name Troyborg or Troytown (and other names indicating

known historical or legendary cities). The alternative use of names for the stone or turf labyrinths

would, to my understanding, relate to a major switch in the power structures in northern Europe

when the kings of the Northern states broke with the supremacy of the papacy. The Reformation

of the Catholic Church, taking place in Northern Europe almost simultaneously in the 1520-40s,

meant that the king would now be the head of the Lutheran and Protestant Churches, and instead

of being elected, their sons would now be hereditary princes to the crown.


A Jerusalem appellation would likely indicate a Christian Catholic use of the labyrinth, laid out

on the ground in a classical- type layout of stones, or on the Continent and in England, more

often in turf, due to a lack of stones as building material. In Ireland, the Hollywood stone, a

boulder incised with a labyrinth of classical type was found on the pilgrim path to Glendalough

Abbey, has been described as a pilgrim route marker by Dr Rachel Moss of Trinity College,

Dublin, in a personal conversation with the author in 2019: “With regard to the Hollywood stone,

it marked the start of the pilgrim route across mountainous terrain to Glendalough. Pilgrimage

is recorded here from the 10th century, but flourished from the early 13

th. However, earlier carved

stones were sometimes repurposed by the Church in Ireland, so this does not necessarily help

with dating.” The Hollywood stone would then connect use of the classical-type labyrinth to

medieval Catholic pilgrimage as an applied concept.

When constructed in turf, the labyrinths required regular maintenance, and knowledge of some

only remains as folklore and legends of their use (e.g. for Easter celebrations and various Catholic

Guild festivals, initiation rites for the Teutonic Order, etc.), or as historical drawings recording

their structure and form, and significantly, surviving as place names on old maps.

Jerusalem place names would then, in a Christian Catholic context such as along known pilgrim

routes, argue in favour of the Catholic Church’s relation to and use of the labyrinth symbol, and

also situate it during the time of medieval pilgrimages. The when and where of the pilgrimages

are supported by various known and dated Nordic itineraries, e.g. the Adam of Bremen sailing

itinerary from Ribe in Denmark, via Plymouth in England to Acre in Palestine from the late 11th

century (figure 1), and the itinerary of the monk Nikolás Bergson from Iceland that records

continental pilgrim routes from the middle of the 12th century. Pilgrim badges from various

continental and Scandinavian destinations have been found in Scandinavian burial grounds, and

some badges moulded into church bells have allowed dating [Andersson 1989]. Scandinavian


pilgrims and their dates of arrival have been identified in monastery hostel guestbooks, e.g., the

Benedictine monastery of Reichenau [Harrison 2020, 79]. We know some of the pilgrim routes,

with pilgrims going south and back north again, continued to be used after the Scandinavian

Reformation of the 1520-40s. It is

presumed that the pilgrim route

guiding marks were labyrinth

constructions called Jerusalem – see

my article “Jerusalem Place-names and

the Baltic Labyrinths” in Caerdroia 49

(2020), p. 42-48.

Fig. 1: (A) Adam of Bremen’s sailing

itinerary from Ribe in Denmark to Acre

in Palestine from the latter part of the

11th century. (B-E) Routes to Santiago de

Compostela from the itinerary in Liber

Santi Jacobi, early 12th century


A labyrinth construction called Troyborg would then associate with an entirely different cultural

heritage. The word Troyborg has a first element of troy- or troj- in Scandinavian languages, and a

second of borg, translated as castle, town or stronghold. Studying the reports of the pioneering

Swedish antiquarian Johan Hadorph, active between 1666 and 1693, John Kraft notes that in

older maps and other written records from the 17th and 18

th centuries, “the labyrinths in the

Nordic countries were as a rule not called ‘labyrinths’ before the 19th century. The old common

names used in Scandinavia were Trojeborg, Trojaborg, Trojenborg, etc., all names that allude to

the ancient city of Troy.” [Kraft 2017, 9]. The earliest of the official reports from local priests

mentioning a labyrinth, handed to Johan Hadorph in 1672 from the parish of Låssa, refers to the

Rösaring labyrinth, situated on high ground above the water fairway leading to the religious

centre of Uppsala to the north and to Stockholm and the Baltic Sea to the east.

John Kraft maintains that the Troy- names would allude to the legendary city of Troy, known

from Homer’s epic poem the Iliad which is believed to have been written down in the 8th


BCE. Troy would also appear in various writings during the Roman period, such as Virgil’s widely

read epic poem the Aeneid, written (but never finished) between 29-19 BCE, and as we shall see,

also in paraphrase writings of the Trojan saga during the Middle Ages.

The Aeneid is a poetic narrative that considered the Roman origin myth, with the Trojan hero

Aeneas presented as the founding father of Rome and progenitor of the Romans. Virgil also

refers to the legends of the labyrinth constructed by Daedalus at Knossos on Crete, that appear

in the same context as the equestrian games of lusus troie (described in Aeneid V:545-605): “As

once in high Crete, it is said, the Labyrinth held a path woven with blind walls, and a bewildering

work of craft with a thousand ways, where the tokens of the trail were broken by the

indiscoverable and irretraceable maze: even in such a course do the sons of Troy entangle their

steps, weaving in sport their flight and conflict, like dolphins that, swimming through the wet

main, cleave the Carpathian or Libyan seas and play amid the waves.” [Virgil, V:588-595]


Of note here should be “the sons of Troy,” young Romans who were trained in equestrian games

as a parade and para-military exercises to commemorate their Trojan ‘ancestors’ and heritage, as

the games would have been performed at major state events, imperial funerals, temple

dedications settlement foundations and to celebrate military triumphs. [Encyclopedia of Ancient

History, s.v Lusus Troie]

The legend of Daedalus’ labyrinthine construction hiding the Minotaur is not taken from the

Iliad, in which Homer is only indirectly relating to the dance of celebration Ariadne would have

performed at Naxos. Theseus slaughtering the Minotaur as an expression of a legendary victory

over death and evil, is not mentioned in the Iliad, so therefore we could assume that the Theseus-

legend was not well known in the 8th

century BCE. It would then seem that the labyrinth concept

is incongruous with reference to the Iliad only. In the Aeneid, however, Virgil’s story of Daedalus

(taken from various sources) and his legendary construction of the labyrinth is made explicit in

book VI, starting at line 14 and continuing to tell of “…the mongrel breed of the Minotaur, a

hybrid offspring, record of a monstrous love; there that house of toil, a maze inextricable; but

Daedalus pitying the princess’ great love, himself unwound the deceptive tangle of the palace,

guiding blind feet with the thread.” [Virgil, VI: 26-30]

If these are references to labyrinths in texts, visual

references to the Theseus and Minotaur legend

could in Roman times also be seen in labyrinth

mosaics. But the labyrinth layout would be the

creation of artists, traditionally dividing the mosaic

labyrinth into four sections, set within walls and

gateways, much like a Roman castrum, with a square

format differing from the classical-type (figure 2).

While labyrinths of the square Roman format might

seem more likely to be the prototype for the

Troyborgs of the North, the question remains as to

why the northern labyrinths are of mainly classical-


Fig 2: a typical Roman mosaic labyrinth,

Cremona, Italy. Photo: Jeff Saward

These references to a labyrinth constructed by Daedalus and descriptions of the lusus troie in the

Aeneid, are – in brief – understood to be the source of the Nordic labyrinth naming of Troyborg

and also Julian’s Bower. Iulus/Julian was the son of the main character Aeneas, who leaves Troy

at its downfall, taking his son by the hand and his father Anchises on his back, setting out for what

eventually would be the founding of Rome. Again, a focus on ancestry leading back to legendary

Troy and its heroes.

There are several interpretations of the connection of legendary Troy and labyrinths, but

questions arise: why would these Troyborg/Julian’s Bower labyrinth structures of identical

classical-type have been placed along known pilgrim routes inland and along coasts and skerries,

stretching from Iceland in the west, via the Baltic Sea to the White Sea in in the east? What was

their purpose and the motivation for building them? And how was the visual tradition of the

classical-type labyrinth transferred and applied to these structures?


In contextual archaeology [Hodder 1986] the meaning of an archaeological artifact, a monument

or effigy, such as labyrinths laid out in stone, painted as frescos, scribbled as graffiti or engraved,

could only be understood in relation to their cultural context. “Only through the study of the local

and specific cultural historical context could we understand what significance a particular artefact

or monument had in the past. The reason being the connections that exist between the society

and human activity on the one hand and the material culture on the other, are totally dependent

on actions of individuals within specific cultural historical contexts. Only through a thorough

reconstruction of the specific cultural historic context, could we hope to be able to understand

the meaning behind the material culture – and ultimately the thoughts of the people in the past”

[Jensen & Karlsson 2001, 56-57, author’s translation]. With the intention to elucidate why and

when the labyrinths were used around the Baltic Sea, a thorough reconstruction of a specific

cultural historical context of the time of the Nordic crusades, starting in the early 13th century,

has provided results.

To start with Denmark, the previously

mentioned early medieval pilgrim route

itineraries credited to Adam of Bremen

and Nikolás Bergson, and the Erhard

Etzlaubs Romweg map from the 1490s,

passes through the Danish mainland of

Jutland. There are no traces left of turf

labyrinths in the case of Denmark, but

Troyborg (and similar) place-names on old

maps have been reconstructed by John

Kraft (figure 3). Fortified constructions

called Trelleborg have been known since

the Viking Age in Denmark [Harrison

2020, 97], but the Troy place-names

correspond in several instances to known

pilgrim routes, e.g. Ribe to Flensburg.

Fig. 3: Map of Troy- place-names in

Denmark. Source: Kraft 1986, in

Westerdahl 2016, fig.19a, p.37

● - Labyrinths with known names

■ - Place names sometimes referred to as labyrinths

or castle ruins, or with no known remains

An illustrative connection between classical labyrinths and the Catholic Church can be seen in

frescos in ten Danish medieval churches. These are in some cases dated and, as my research has

demonstrated, also connects to moulded pilgrim badges on the dated bells of some of the same

churches, e.g. at Hesselager. The pilgrim badges would then connect the labyrinth symbol to the

medieval pilgrimages. The dating of these frescos and pilgrim badges to the latter part of the 15th

century provides in archaeological terms a terminus ante quem that labyrinths would occur in

medieval Catholic church contexts, i.e., the concept would have been part of this cultural context

prior to that time. Similar circ*mstances can also be shown in a few important cases in Sweden

(Gotland) and the relationship of medieval church frescos and place names in Finland. The name

replacement hypothesis would explain why Finland would either keep Jerusalem or call their

labyrinths Jungfrudanser – the Virgin (Maria?) dances.


The cultural context would by implication include a social aspect of the labyrinth in practice. The

labyrinth paintings would manifest a Catholic religious symbolism that the congregation would

understand. The Black Death would hit the countries of the North with full force ca. 1349-52 with

a loss of probably more than half of their populations [Harrison 2018, 74]. As the dating of the

labyrinth paintings and the moulded pilgrim badges in church bells is of a post-pestilence period

by more than a hundred years, this would imply the long-time struggle of the church with such

heavy losses and the need for revival. As the earlier years of medieval pilgrimages were halted,

we may assume that the concept of the classical labyrinth, and in Denmark possibly laid out in

turf, was the original and familiar Christian labyrinth symbol.

In earlier labyrinth research [Kern, Saward, Wright, Reed Doob, etc.] of the origin, development,

distribution and use of the visual labyrinth symbol in Europe, the author finds convincing

arguments for a medieval Christian Catholic use of the labyrinth symbol in Europe. These

symbols as imagery; in manuscripts, calendars and computus, effigies in churches and as roadside

markers, inlaid cathedral pavements, as well as their metaphorical use in written accounts such

as Dante’s Commedia, seem to correlate in time with the pilgrimages starting in the 11th century.

Initiated with focus on the Holy Land and in the North to St Olav’s tomb in Norwegian

Trondheim – and continuing through the Middle Ages, with the city of Jerusalem closed, turning

instead to Santiago de Compostela in Galicia – the pilgrimages flourished in the 13th

and 14th

centuries. With focus on Rome starting the jubilee year 1300, the year Dante makes his pilgrimage

in literature in the Divine Comedy – the movement was halted with the pestilence years starting

around 1350. Pilgrimages would later continue in the North closer to home up to the time of the

Reformation, during the first half of the 1500s.

My suggestion is that there has been a shift of names, and hence the use of the labyrinths, from

Jerusalem to Troyborg along with the Reformation in the Nordic countries and England in the

1520-40s. The radical Reformation of Denmark, starting in the trade and harbour town of Malmö

in the county of Skåne, at the time within Denmark is summarised: “The year 1536 the parliament

in Copenhagen formally put into effect the Reformation in Denmark, which would bring about

the end of the bishop’s and church’s political power. Henceforward the king would be the head

of the church. The reform made fast progress, strongly influenced by economic factors… As a

result of the Reformation’s critical studies of the Bible, many of the Church’s customs such as

pilgrimages, priest celibacy, relics and the wealth of the Church, were questioned.”

[, author’s translation]

In England, the Reformation is said to start with Henry VIII’s quest for a male heir. In 1534 the

king declared that he alone should be the final authority in matters relating to the English church.

Ancestry of the Crown and its heirs would hence be of vital importance. In Sweden, the

Reformation is said to have been put into effect in the riksdag of Västerås in 1527 and in 1544

[]. At the latter riksdag, the so-called Succession Parliament allowed King

Gustav Vasa I to make the succession of the throne hereditary. The king was then not only head

of the church, making the Crown able to confiscate the property of the Church to pay his wartime

debts, but also made his sons hereditary princes. The suggestion is that a narrative to confirm this

new power structure of a hereditary Crown and head of church, claiming lineage back to a

legendary heroic origin, would find expression in the Trojan saga, a paraphrase on the Iliad,

complementing the study of Virgil’s Aeneid in schools continuously since Roman times up to the

1900s. [Flower Smith 1916]


It has been suggested above that there could have been a forced replacement of labyrinth names

from Jerusalem to Troyborg or Troytown in Scandinavia and England due to the Reformation.

There was no longer reason for pilgrimages and symbols for Jerusalem but the labyrinths were

supposedly there and still probably considered as powerful in their social cultural contexts.

Strategically set as they were, to have been seen and venerated by many “to show the way,” a

transformed symbolism of the labyrinths could be used as seals of propaganda for the

Reformation and the king as new head of the church. The concept of the Trojan saga would then

provide a narrative and alibi to also level with the congregations for the new order, encouraged

by the Crown. As Sweden, Denmark and England share the historical sequence and prerequisites

for the Reformation around the 1520-40s, the Historia Trojana may have played a potent part, a

perfect legendary fit together with versions of Virgil’s Aeneid.

Historical novels and versions of the Historia Trojana, a paraphrase of the Iliad, and hence

without the references to Roman ancestry or to Daedalus, circulated in Europe during the Middle

Ages. The most famous version is the verse novel by Benoit de Sainte More from ca. 1165. The

legend of Troy seems to have appealed as a legendary origin to the Germanic kings and rulers, as

had been Virgil’s Aeneid to the Roman emperor Augustus, for its origin myth of Roman descent

from the legendary city of Troy. At the time of the Reformation in the first half of the 16th

century, the Historia Trojana was copied into vernacular languages, including Swedish in 1529

[NE, s.v. Trojasagan]. The Historia Trojana and its descendants may then have initiated a popular

legendary history around Europe during the Middle Ages. In 1996, Michael Behrend wrote:

“Accounts of the Troy romances can easily be found in libraries, so to be brief: the most

influential works were the 12th century French Roman de Troie by Benoît de Sainte More (or

Maure), and a late 13th

century translation of this into Latin, the Historia Trojana by Guido of

Colonna. Guido’s version was immensely popular and was the basis for versions of the story in

German, Italian, English, Scots, French, Spanish, Low Saxon, Dutch, Danish, Flemish and

Bohemian. There was also an Irish version even before Benoît, and an Icelandic Trojumanna

Saga. A French version of Guido, Raoul le Fèvre’s Recueil des Histoires de Troyes, was translated

by William Caxton and published by him in 1474 – the first book printed in English.” [Behrend


And further: “Apart from the popularity of the Troy romances, the Troy legend was important in

the European Middle Ages because people firmly believed that descendants of Trojan refugees

had founded nations in Europe. The story of Brutus, the supposed great-grandson of Aeneas and

founder of the British nation, is told for example in Latin by Nennius (9th century, briefly) and

Geoffrey of Monmouth (ca. 1136, at length), and in English in Layamon’s Brut (late 12th century).

It was still appearing in chapbooks as late as the 18th century.” [Behrend 1996]


In this article, I have tried to reason for the use, distribution and time of diffusion of the Nordic

labyrinths through their commonly applied name of Troyborg (since the 17th

century in Sweden)

in relation to their supposed earlier name of Jerusalem. Tracing the use of classical-type

labyrinths back to the early Middle Ages in Catholic Europe, following their tracks along pilgrim

routes in Ireland, Denmark and in an earlier article [fa*gerström 2020] through today’s Poland

and Germany, as well as their 15th century occurrence as frescos in medieval churches in

Denmark, the labyrinth symbol seems to have had a clear connection to the concept of

pilgrimages. The supposed Catholic name of Jerusalem, known from place names on old maps,

from folklore and legends, and the classical labyrinths found in manuscripts and drawings as well

as in medieval church frescos, would further indicate a time for their use and diffusion.


If the labyrinth symbol up to the time of the Nordic Reformation of the Church had been closely

connected to the Catholic Church and its system of pilgrimages, it would as a seal of the papacy

and the Catholic faith have been considered socially powerful. As worldly manifestations they

could therefore become useful as seals and symbols to confirm the new regal power. It is reasoned

that the traditionally laid out stone or turf labyrinths of earlier classical-type, large enough to

walk or ride, could be associated with the important claims of Trojan origin and legacy of the

Reformation kings, as expressed in the Historia Trojana (complemented by Virgil’s Aeneid and

specific references to Daedalus and the lusus troie). The suggestion is that the stone and turf

labyrinths would by high authority have been given the names of Troyborg in today’s countries of

Denmark and Sweden, and Julian’s Bower in Norway and England. The new names would then

be applied to already existing structures and hence explain why the stone labyrinths would keep

their classical format with a change of name from Jerusalem to Troyborg.

Christina fa*gerström, Nybro, Sweden; March, 2021

Email: [emailprotected]


Andersson, L. “Pilgrimsmärken och vallfart.” Lund Studies in Medieval Archaeology, Lund, 1989.

Behrend, M. “Julian and Troy Names.” Caerdroia 27 (1996), p. 18-23.

The Encyclopedia of Ancient History, s.v. “Lusus Troie.” London: Blackwell Publishing, 2013.

fa*gerström, C. “Jerusalem Place-names and the Baltic Labyrinths.” Caerdroia 49 (2020), p. 42-48.

Flower Smith, K. “The Later Tradition of Virgil.” The Classical Weekly, vol. 9:23 (1916), p. 178-182.

Harrison, D. Digerdöden. Lund: Historiska Media, 2018.

Harrison, D. Sveriges Medeltid. Lund: Historiska Media, 2020.

Hodder, I. Reading the Past. Cambridge University Press, 1986.

Jensen, O. W. and H. Karlsson. Aktuell Samhällsteori och Arkeologi. Göteborg: Bricoleur Press, 2001.

Kern, H. Through the Labyrinth. Munich, London & New York: Prestel, 2000.

Kraft, J. “Built in Honour of Odin and Danced Around.” Caerdroia 46 (2017), p. 9-16.

NE. Nationalencyklopedin, 1995 ed., s.v. “Trojasagan.”

Reed Doob, P. The Idea of the Labyrinth. Ithaca: Cornell University Press, 1990.

Saward, J. Labyrinths & Mazes, London: Gaia Books, 2003.

Virgil. The Aenied.

Wright, C. The Maze and the Warrior. Cambridge, MA: Harvard University Press, 2004.

Electronic references

Danish churches:

Reformation in England:

Reformation in Sweden and Denmark:



Simple Alternating Transit Mazes

Richard Myers Shelton

Abstract: SAT mazes, introduced by Tony Phillips, are a useful class of full-course labyrinths that form

a natural algebraic system. Several families of SAT mazes show up in the real world, and they serve

as the basis for the Roman-style compound labyrinths.

We are all familiar with full-course labyrinths like the Classical labyrinth that have a main axis

but no internal axes. Tony Phillips wrote about an important subclass of these on his website

Through Mazes to Mathematics [Phillips], and this subclass is so basic to the study of more complex

labyrinths that it is worth taking some time to explore his approach.

The class Phillips defines is the collection of simple alternating transit mazes (SAT mazes). Their

courses are idealized as concentric circles. “Simple” means that the path traces each course

completely before moving on to a different course: each segment of the path is a circuit along a

single course, connecting one side of the main axis to the other. “Alternating” means that at each

change of course (necessarily along the main axis in these mazes) the path changes direction

without crossing the axis: successive courses in the path are traced in opposite directions.

“Transit” means that the path leads in a simple non-branching way from the exterior through all

the courses to the interior and then stops: all the courses are completely traversed on the way to

the center and there is no separate return path. So the SAT mazes are just the full-course

labyrinths (1-axis labyrinths) that don’t cross the central axis and whose path stops at the center.

The canonical example of a SAT maze is, of course, the 7-course Classical labyrinth.

Phillips talks in terms of levels rather than courses. Each course is a separate level, but the interior

and the exterior are levels as well. The levels are numbered from 0 for the exterior, 1 for the

outermost course, 2 for the next course in, up to L for the interior. The number of levels in a

labyrinth includes the interior, but not the exterior; for Phillips the 7-course Classical labyrinth is

an 8-level labyrinth, and in general a labyrinth of N courses has L = N + 1 levels (counting the

interior but not the exterior).

It is useful to think of a SAT maze as an action: you start with the universal level 0 (the exterior,

the empty infinite Euclidean plane), and by drawing the labyrinth you impose upon the empty

plane a series of new levels (the courses and the interior): the labyrinth consists of the new levels

1 through L that are imposed upon the exterior plane. Drawing the Classical 7-course labyrinth,

for example, starts from the empty plane and creates 8 new levels: the 7 courses and the interior.

The number of new levels created by the SAT maze is its height. The Classical labyrinth thus has

height 8: it adds 8 new levels to level 0. The height is also the number of concentric walls required

to delineate the new levels: the act of drawing 8 new levels is really the act of drawing 8 walls on

the empty plane. The height also corresponds to the number of connections between levels along

the sides of the main axis.

Labyrinths – especially SAT mazes – are frequently shown in level charts, with their levels unrolled

into horizontal steps rather than circling around the common center.1 Each course is shown as a

horizontal line, representing the path as it travels along that course (Figure 1). The interior and

exterior levels are not represented by separate horizontal lines in the level chart; they are just the

areas above and below the set of lines representing the courses.


Each SAT maze A has a dual, a SAT maze A′ formed by flipping the level chart over from top to

bottom, or by rotating the plane of the chart by 180 degrees. Either operation interchanges the

notion of interior and exterior: the innermost course of A becomes the outermost course of A′,

and vice versa. (The two operations yield results that are mirror images of each other, but we

regard them as equivalent because we don’t distinguish between a labyrinth and its mirror

reflection: mirroring the chart from right to left does not change the relationship between the

courses.) Clearly the dual operator applied twice gets you back where you started: the dual of A′

is A back again, since rotating the level chart by 180 degrees twice restores the original


Some SAT mazes, like the Classical labyrinth, are self-dual: when you flip the level chart over you

get the same chart (or its mirror reflection). Such labyrinths have an inherent sense of balance or


Level Sequences

The ground rules for SAT mazes mean that the maze is completely determined by specifying the

number of levels and the order in which they are visited. Thus, we can make a prototypical

mathematical move by identifying a SAT maze with the sequence of numbers describing that

order. This sequence is called, naturally enough, the level

sequence. The Classical labyrinth corresponds to the

level sequence (0, 3, 2, 1, 4, 7, 6, 5, 8), as shown in

figure 1. Each adjacent pair of numbers in the sequence

implies a connection between those two levels along the

main axis. The pairs starting with an even number

represent connections along the “entrance side” of the

axis, and those starting with an odd number represent a

connection on the other side of the axis.

Fig. 1: The Classical labyrinth as (0, 3, 2, 1, 4, 7, 6, 5, 8)

The key point is that not every sequence works: just as an arbitrary squiggle need not be a SAT

maze, an arbitrary sequence of numbers need not correspond to one: like the squiggle, the

sequence must obey certain rules. Obviously, it must start with 0 and end with L, as the maze

starts at the exterior and ends at the interior; and every number in between (every course) must

appear exactly once. This means that the sequence is just a re-ordering of the numbers from 0 to

L, i.e., the sequence is a permutation of the numbers from 0 to L. In addition, as Rule VI of

[Shelton 2019] shows, the numbers must alternate in parity, switching between even and odd: an

odd-numbered level can connect only to an even-numbered level, and vice-versa. (Otherwise the

portion of the path falling between those two levels will get boxed in.)

Finally, the order of the numbers in the sequence must capture the notion that the path doesn’t

cross itself. This translates into conditions on pairs of adjacent numbers in the sequence. The

pairs beginning with even numbers represent connections along the entrance side of the axis, and

these must not cross each other: thus if the sequence includes pairs a, b and m, n (where a and m

are even and b and n odd), then either the turn represented by m, n is contained inside a, b (i.e.,

both m and n lie numerically between a and b); or a and b must lie numerically between m and n;

or the two turns are completely separate (the numerical ranges a-to-b and m-to-n don’t overlap).

And the same restriction must apply to connections on the other side of the axis as well, to pairs

starting with an odd number.


This restriction is easy to verify in Figure 1 for the Classical labyrinth. For example, the pair 0, 3

at the beginning of the level sequence numerically encloses the pair 2, 1; but it is numerically

disjoint from the other two pairs that start with even numbers (4, 7 and 6, 5). Likewise, on the

other side of the axis, 3, 2 is enclosed by 1, 4; and both are disjoint from 7, 6 and 5, 8.

Any SAT maze automatically generates a level sequence that obeys these rules. Conversely, any

level sequence that obeys these rules does correspond to a SAT maze: you can start with a stack

of L−1 horizontal lines, and then connect them in the order specified by the sequence – and the

rules guarantee that you will end up with a level chart whose path does not cross itself or the main

axis. From this level chart you can then draw the corresponding SAT maze. This construction

therefore yields a one-to-one correspondence between the rule-obeying level sequences and the

set of all SAT mazes. The sequences thus provide a mathematical model for SAT mazes.

The representation of a SAT maze as a level sequence emphasizes that the SAT maze is

characterized completely by (1) its height, and (2) the order in which the courses are traversed.

Other differences are immaterial: it doesn’t matter, say, whether the shape of the courses is

precisely circular, or square, or whatever. Nor does it matter which side of the axis the entrance

is on: a labyrinth and its mirror image are considered two examples of the same underlying type.

Phillips proceeds to relate such sequences to other phenomena (such as the ways a perforated

strip of postage stamps can be folded into a stack), and to look at ways of counting the number

of such sequences; but that’s a direction I don’t want to pursue here. What does interest me is the

way Phillips builds composite labyrinths by stacking smaller components and connecting them


Composition of SAT Mazes

Phillips introduces the operation of composition, represented by an asterisk (*). Composition

joins two SAT mazes into a larger one. If A and B are SAT mazes, then A * B is the SAT maze

you get by drawing B inside A, or equivalently, by stacking the level chart of B on top of the level

chart of A (figure 2).

The right way to think of this is that you are drawing a copy of B inside the interior of A: the

interior of A becomes the exterior for B, and the new levels of B are imposed upon the interior

of A. Thus the levels of the composite (the levels imposed on level 0 by drawing A * B) are the

courses and interior of A, followed by the courses and interior of B. The interior of A overlaps

with the exterior of B, forming a new course separating the courses of A from the courses of B.

When you draw B inside A, you have to start B with its entrance on the opposite side from the

exit of A, so that the new course connecting them does form a full course. This means that the

copy of B inside A may have to be the mirror image of your original B.

Fig. 2: Composition A * B of mazes A and B:

left to right: (a) Maze A: 4 courses, height 5

(b) Maze B: 2 courses, height 3

(c) B drawn (mirrored) inside the interior of A: 7 courses, height 8


The convention of counting the interior (but not the exterior) as a level of the labyrinth makes

the numbers work out right: if A has L levels (including the interior) and B has M levels (including

the interior), then in the composite, the count L includes the courses of A and the new course

between A and B, and M includes the courses of B and the interior. Thus the composite

conveniently has L + M levels (which, again, includes the interior).

The level sequence of A * B is roughly just the level sequence

of A followed by the level sequence of B – except that you have

to add L (the height of A) to all the numbers in the second

sequence, because in the composite the levels in the second

component start from L instead of 0. The number L thus

appears at the end of the first sequence and the beginning of

the second, and this is where the two levels overlap: to join the

sequences, you combine the two Ls into a single L, which

represents the new course between the two components.

Fig. 3: The Classical maze as the composite of two meanders

It is easy to recognize whether a SAT maze is a composite: there will be a single course that

divides the maze into two pieces: one part of the maze (including its connections along the axis)

lies below this course and the rest lies above it. The Classical labyrinth, for example, is the

composition of two meanders joined at level 4 (figure 3). In terms of the level sequences we have:

(0, 3, 2, 1, 4) * (0, 3, 2, 1, 4) = (0, 3, 2, 1, 4, 7, 6, 5, 8)

where we’ve added 4 to all the numbers in the second sequence and overlapped the terminal 4 in

the first with the initial 4 in the second.

Just as with multiplication of ordinary numbers, the composition operator * can be omitted if the

context is clear: AB is written for A * B. Likewise A * A can be written A2, and similarly for higher

numbers of repeated composition. Unlike multiplication, however, order matters in composition:

B * A is stacked in the opposite order from A * B, so unless A and B are two copies of the same

maze, the two orders typically yield different labyrinths.

Families of SAT Mazes

The standard Roman-style labyrinth takes a SAT maze (the base) and replicates it (usually four

times) around a common center, linking the base mazes with two extra courses that connect the

exit of one base to the entrance of the next. In his survey of surviving Roman mosaic labyrinths

[Phillips 1992], Phillips introduces and labels the various SAT mazes that appear as the bases.

These fall naturally into families, and Phillips assigns a Greek letter name to each family, and

distinguishes one member of a family from other family members by adding a subscript to indicate

the number of levels it has. In the following, I will highlight the families of Phillips, together with

a few more that pop up with some frequency.

Epsilon and the Serpentines (εN

). Phillips introduces the letter epsilon (ε) for the trivial labyrinth,

which is about the simplest SAT maze: it has one level (the interior – as usual we don’t count the

exterior) and no courses (figure 4). As a labyrinth it is simply a circular wall separating the interior

from the exterior, with a gap to allow the path to enter. The path itself is a simple straight line

from the exterior to the interior. Epsilon is the only labyrinth whose path doesn’t turn to the right

or to the left as it leaves the exterior, so it is the only labyrinth with a path that is left-right mirror



Fig. 4: Serpentines – powers of epsilon

Epsilon is useful in building new

labyrinths from old. Composing any SAT

maze with ε simply adds a new course at

the top or bottom: (A * ε) is A with a new

course at the top to connect A with ε, and similarly (ε * A) is A with a new course at the bottom.

In particular ε * ε = ε2 is the single-course labyrinth: two trivial labyrinths connected by an added


We can extend this indefinitely. If you compose the single-course labyrinth with another ε, you

get a two-course labyrinth; and continued composition with ε leads to labyrinths with more and

more levels that snake back and forth in serpentine fashion. εN, the composition of N copies of ε,

is the N-level or (N−1)-course serpentine or back-and-forth labyrinth. If we introduce subscripts

to distinguish the various members of the serpentine family, we get a very simple relationship:


, the (N−1)-serpentine with N levels (or N−1 courses), is just the N-th power of the trivial

labyrinth ε under composition.


= εN = ε * ε * … * ε (N times)

In particular, ε itself belongs to this family: ε = ε1 = ε


, even though the trivial labyrinth doesn’t

look much like a serpentine: it’s the serpentine with 1 level and therefore 1−1 = 0 courses.

It is easy to smile at the serpentine labyrinths – they seem so simple – but they have a long history.

They are the SAT version of spirals: moving inward one course at a time, but not crossing the

axis. As we shall see, the 3-course Classical labyrinth is just the 3-serpentine ε4. In the real world

there are several walkable serpentines with more levels. The labyrinth in Aspang, Austria

[Lindenmayr] is the serpentine ε9 with 8 courses, and the Katzenlabyrinth of Monika Bugs in

Saarlouis, Germany [Bugs], with 4 courses, is an elaborately decorated version of ε5.

Meanders (γN

). The gamma family (figure 5) are the simple meanders of increasing size: γN


N levels and N – 1 courses, where N is always even. They are formed by making “maximal jumps”:

from the exterior to the highest odd course, then to the lowest even course not yet visited, then

to the highest odd course not yet visited, and so on.

The first maze in this family is γ2

, the single-course labyrinth. As we have seen, this is a composite:

two copies of ε separated by the single course, so γ2

= ε * ε = ε2. Unsurprisingly therefore, the

powers of γ2

are serpentines: specifically, the serpentines with an odd number of courses:


)N = (ε2

)N = ε2N

, the (2N−1)-serpentine. The even serpentines are just the odd serpentines

with an added course: ε2N + 1 = γ


N * ε.

Fig. 5: Meanders – γN

(The expression γM

N should be read (γ


)N – the exponent goes with the labyrinth, not with its

subscript: “N copies of γM

”, not “a single γ with M N levels”. The number of levels in a power is

the exponent times the subscript, and the number of courses is one less.)


The larger gammas, however, are not composite: they do not fall into two disjoint pieces

separated by a single course, as the jump from the exterior to the highest course encloses all of

the other courses.

Reverse Serpentines (βN

). The beta family are the reverse serpentines: the path goes immediately

to the innermost course and then backs out in serpentine fashion before heading from the

outermost course to the interior (figure 6). Phillips did not name this family, since the only

examples occurring in Roman labyrinths are β2

and β4

, which are the same as γ2

and γ4

. But I have

encountered other betas often enough to have introduced a name for them, following the pattern

set by Phillips. (I chose beta since the letter’s shape suggests the initial run to the top of the

pattern.) As with the gammas, β2

is the

single-course SAT maze, and is therefore

composite; but in larger cases the

enclosing entrance and exit connections

make them non-composite.

Fig. 6: Reverse serpentines – βN

Classical vs Otfrid labyrinths. The gammas are important because they appear in the Classical

labyrinths. γ4

appears twice in the 7-course Classical, which (as we’ve seen above) is the composite


2 = γ


* γ4

, i.e., two 3-meanders connected by a single course; and it has (as we would expect)

4 + 4 = 8 levels, or 7 courses.

There are two natural ways to extend the Classical labyrinth (figure 7). You can make the

component meanders bigger, as in the 11-course Classical labyrinth, which is γ6

* γ6

with 12 levels;

or you can stack more copies of γ4

on top, as in Otfrid, which is γ4

3 = γ


* γ4

* γ4

(also with 12

levels). Thus two natural sequences include the Classical labyrinth: the Classical sequence (2

copies of bigger and bigger meanders) and the Otfrid sequence (more and more copies of the 3-

meander γ4


At the small end of these two sequences stand the 3-serpentine γ2

2 and the 3-meander γ



Although the 3-meander is often called “the 3-course Classical labyrinth”, the correct version

(the one built on analogy with the 7-course and 11-

course Classicals, and the one corresponding to the

seed pattern with a cross and 4 dots) is the 3-

serpentine, not the 3-meander.

The Classicals are so important historically that I

give them their own letter, kappa (κ), representing

the initial k sound in “Classical”:


= γ2

* γ2

= γ2

2 : Classical-3 with 4 levels;


= γ4

* γ4

= γ4

2 : Classical-7 with 8 levels;


= γ6

* γ6

= γ6

2 : Classical-11 with 12 levels;

and so forth.

Fig. 7: The Classical sequence (top)

and the Otfrid sequence (bottom)


Mixed Classicals. The Classical sequence (Classical-3, Classical-7, Classical-11, etc.) grows by

jumps of 4: each time you increase the size of the component meanders, you add two levels to

each meander, and thus 4 to the composite. But what about the intervening numbers, like

“Classical-5” or “Classical-9”? These would have 6 and 10 levels, respectively – and while 6 and

10 are even numbers, dividing them by 2 yields odd numbers, whereas the gammas only come

with an even number of levels. To assemble the 5-course Classical labyrinth, we can’t use two

copies of the non-existent γ3

, so instead we pair together γ2

and γ4

. But there are two ways to do

this: (γ2

* γ4

) and (γ4

* γ2

). So there are two distinct versions of “the” 5-course Classical labyrinth

(figure 8); and similarly, other intermediate steps like 9, 13, or 17 courses can only be

approximated by versions whose halves are close but not identical in size.

The Classical labyrinths are self-dual, but the mixed Classicals are not, because the two halves are

of different size. However, the two mixed Classicals of the same height are duals of each other.

This is because, in general, the dual of (A * B) is the dual of the two components in reverse order:

(A * B)′ = B′ * A′

(Intuitively, flipping the level chart of A * B flips each of A and B, but stacks the flipped

components in reverse order.) Since each meander γN

is itself self-dual, we have, for example:


* γ6

)′ = γ6

′ * γ4

′ = γ6

* γ4

Only one Roman labyrinth uses

an “intermediate” or “mixed”

Classical as its base: the 13-course

maze γ6


appears in Giannutri

(Kern 139).

Fig. 8: Mixed Classicals

Baltics (δN

). The traditional Baltic labyrinth is a simple meander that typically comes with a spiral

into the center and an exit path from the center to the outside. From the viewpoint of SAT mazes,

the extra spiral and exit path are “inessential variations”, and if we leave them out, what’s left is

the meander (a member of the γ family) with an extra course on the inside. I give this family the

Greek letter delta (δ); the extra stroke at the top of the delta serves as a mnemonic for the extra

course at the top. These SAT versions of the Baltics have an even number 2N of courses, and are

formed simply by adding an additional course at the top of the corresponding gamma: Baltic-2N

= δ2N + 1

= γ2N

* ε. Thus Jeff Saward’s “St. Patrick’s Purgatory”, stripped of inessentials, is Balitic-

4 = δ5

= γ4

* ε, and the vanished extravaganza at Stolp (Słupsk) in Poland was Baltic-14 = δ15



* ε. Baltic-2N can be viewed as a 2-course loop folded into N layers or plies (figure 9).

Fig. 9: Baltics as folded loops


Reverse Classicals (αN

). The alpha family (figure 10) are the reverse classicals: they jump from

the entrance straight to the top and trace two meanders downwards, much like the betas jump to

the top and trace serpentines from top to bottom. Phillips calls these “double meanders”, but that

phrase more aptly describes the Classical kappas; so, given the paradigm of the betas, I prefer the

name “reverse Classicals” for the alphas. Unlike the Classical labyrinths, which unite two

meanders in a composite, the alphas are not composite (except α2

), for the same reason that the

betas are not composite: their entrance and exit connections along the main axis enclose both of

the meanders and prevent the whole from falling into two disjoint pieces.

Unlike the subscript for the gammas and the betas, N here is not just any even number: it must

have form 2 + some multiple of 4, since each step grows by 4 courses (because each component

meander grows by 2). The first two

alphas duplicate labyrinths we’ve

already seen: α2

= γ2

= β2

= ε2


single-course labyrinth), and α6

= β6

(the 5-course reverse serpentine). But

thereafter they start to be more


Fig. 10: Reverse Classicals – αN

Phillips defined the alphas primarily to account for the elaborate Roman labyrinth at Souse (Kern

169), whose 4 quarters are copies of α10

2. This base has 2 × 10 = 20 levels, or 19 courses, so the

whole assembly (with the 2 extra courses connecting the four quarters) has 21 courses (figure 11).

This is the only surviving regular Roman labyrinth whose base is not one of the gammas or a

composite of gammas.

An interesting relationship connects α6

with the Chartres family. Just as dropping

the internal turns from Greys Court yields

the Classical labyrinth [Shelton 2010],

dropping the internal turns from Inner

Chartres yields α6

– and Chartres and

Saffron Walden (which are essentially 2

and 3 stacked copies of Inner Chartres)

similarly become α6

2 and α



Figure 11: Sousse = 4 × α10


Key: White = γ4

meanders within the α10


Light shade = the connections between quadrants;

Medium shade = connections between meanders within the α10


Dark shade = connections uniting two α10

components into the α10

2 base labyrinth.


Snakes (ηN

). The eta family (figure 12) is an elaboration on the serpentines, so I use the letter

eta (η) for them, as eta is something of an elaboration of epsilon. There are two varieties of

Snakes: the Descending Snakes (the etas: ηN

) and the Ascending Snakes (the duals of the etas:


′ ). They are formed by interweaving an

ascending serpentine with a descending

serpentine, and together these form a

snake-like figure that is easily visible in the

level charts. The etas start with the

descending serpentine; their duals start

with the ascending serpentine. The two

interwoven serpentines share the same

number of courses, so the total number of

courses is even. The height N is thus

always odd.

Fig. 12: Snakes – ηN

and ηN

Et cetera. Of course, there are many SAT mazes that do not belong to these families. The

traditional Jericho labyrinth (figure 13), for example, was modified from the Classical labyrinth

to conform to the tradition that Jericho had seven walls and thus only six courses. Its level

sequence is (0, 3, 4, 5, 2, 1, 6, 7). It looks like Baltic-6 on one side of the axis; but it tries to preserve

something of the structure of the Classical labyrinth on the entrance side, and so maintains the

connection from the exterior to course 3, instead of moving it to course 5 as in Baltic-6.

Fig. 13: Jericho Fig. 14: Tal

The 9-course labyrinth in the mosque at Tal, Pakistan [Saward 2003, 60], looks at first glance like

one of the mixed Classicals (figure 14), but on closer inspection its level sequence turns out to be

(0, 5, 4, 3, 2, 1, 6, 9, 8, 7, 10). This works out to be the composite α6

* γ4

(rather than the mixed

Classical γ6

* γ4

). It’s the only real-life example I know of that mixes the alphas and the gammas.

But another way to analyze Tal comes from observing that α6

= β6

and γ4

= β4

. Thus Tal can also

be written as β6

* β4

, the composition of two reverse serpentines. This shows that Tal just misses

being β6

* β6

= α6

* α6

= α6

2 (Chartres stripped of internal turns).


The Null Labyrinth

Although the trivial labyrinth ε = (0, 1) at first seems fairly strange, there is another labyrinth

that is even stranger, namely the SAT maze that corresponds to an even simpler level sequence:


This is a perfectly good permutation. It’s a permutation of the numbers from 0 to L where L

equals 0, and it satisfies all the rules: it starts with 0 and ends with L; each additional number in

the sequence changes parity (because there are no additional numbers to violate that rule); and

none of the pairs in the sequence breaks the nesting rules (because there are no pairs at all). Each

number in the level sequence corresponds to a level, so this sequence has only one level. Since 0

is always the exterior, this one level is the exterior. Since the highest number in the sequence

corresponds to the interior, level 0 is also the interior. Since the exterior and the interior are the

same, there is nothing – not even a path of zero courses – between them. Since the height is always

numerically equal to the level of the interior, the height of this labyrinth is 0.

This is the null labyrinth, the labyrinth you get by not drawing anything. Since you’re not drawing

anything, nothing divides the exterior from the interior – so they remain the same area, namely

the entire plane. It’s not immediately clear how many courses this labyrinth has. You might guess

0, or perhaps better L−1 = −1; but the correct answer is “undefined” – you haven’t drawn

anything, so there is no path, and therefore the notion of a course for it to follow makes no sense.

So, yes, the null labyrinth is a bit arcane. So why bother with it? For a mathematician, there are

two good reasons. The first I’ve already mentioned: it gives something for the perfectly valid, rule-

obeying level sequence (0) to refer to, and mathematically that’s an excellent reason: it means

that the correspondence between SAT mazes and well-behaved level sequences has no funny


But the second reason might be more convincing. Suppose we have some SAT maze, call it A.

What happens if we draw the null maze inside the interior of A – by not drawing anything? What

we get (obviously) is just A, the maze we started with. Or suppose we take the interior of the null

labyrinth – which is the same as its exterior, which is the whole plane – and draw A inside it. Again,

we get just A. In other words, if we call the null labyrinth O, this shows that

A * O = A and O * A = A

This means that the null labyrinth is the identity element for the composition operator: O plays the

same role for composition of SAT mazes that 0 does for addition of numbers or that 1 does for

multiplication. When one of the operands is the identity element, the result is just the other

operand back again. This gives the algebraic system formed by the set of SAT mazes with the

operation of composition some nice algebraic properties. Mathematicians call such a system a

semigroup with identity (or a monoid), and the algebraic behavior of such systems is a topic in the

field of mathematics called Abstract Algebra.

In our algebraic system of SAT mazes, we have seen that it makes sense to talk about composing

a maze with itself multiple times, and we can use exponential notation for that. So we have, for

example, the Otfrid series of multiple meanders:


( = γ4

1), γ


2, γ


3, γ


4, …

This is similar in construction to exponentiation of an ordinary number multiplied by itself


x ( = x1), x

2, x

3, x

4, …


With ordinary numbers, we are used to extending the exponential notion to allow the exponent

0, which signifies the product of “0 copies of x.” We can do this because multiplication of ordinary

numbers has an identity element, namely 1, and the “empty product” by definition has a value

equal to this identity element. This makes the addition rule for exponents work out right, so that,

for example,

1 ∙ x3 = x

0 ∙ x

3 = x

(0 + 3) = x


Similarly, since the null labyrinth is the identity element for the composition operator *, we can

extend the exponential notation in the system of SAT mazes to include exponentiation by 0: for

any SAT maze A, we define A0 to be the null labyrinth. In particular, since we have already shown

that εn

is εn for positive values of n, we can extend this to include n = 0 to represent “0 copies of

ε”, and get thereby a handy symbol for the null labyrinth:

Define ε0

to be the null labyrinth = ε0

The subscript works out right, too, since the null labyrinth does have height 0 (and is the only

labyrinth of height 0).

SAT Mazes in Roman Composites

Roman labyrinths are largely covered in Chapter VI of [Kern 2000]. Kern’s collection comes

principally from [Daszewski 1977] and adds Side, the Palatine “fountain” at Domus Flavia in

Rome, Cirencester (two examples), Saint-Cyr-sur-Mer, Mieza, Salinas de Rosio, and various

fragments and centers. Daszewski also includes San Vitale (Ravenna), which Kern does not

include in Chapter VI as its inlaid labyrinth dates from the Renaissance, not from Classical times.

Of the 61 pictured Roman items in Kern’s Chapter VI, 44 are fairly standard Roman compounds

built on SAT bases. Of the remainder, 10 are not labyrinths or are too fragmentary to be classified

by type, 3 are simple Classical labyrinths, 3 (Pula, Sparta, Mieza) are in the Roman style but don’t

follow the regular formula, and 1 (Domus Flavia, Kern 115) is almost certainly modern. To these

totals we can add:

• Two square simple Classicals, one at Coimbra, Portugal (adjacent to Kern 130 in the House

of Fountains) and one on a votive pillar from Smira, Kosovo.

• A simple 5-meander γ6

at Halstock in Dorset.

• A standard Roman compound of four Classicals at Mérida, Spain.

• A standard Roman compound at Huete, Spain. (This is very fragmentary, but the base is a

large member of the Otfrid sequence, probably γ4


• A standard Roman compound in the Baths of Julia Memmia at Bulla Regia in Tunisia.

(Also fragmentary, but the base is a large serpentine, probably γ2


Of those of Classical date whose type can be determined, that leaves 47 more or less standard

Roman-style compound labyrinths with SAT bases.2

I say “more or less standard” because the survivors are not all completely regular. Phillips argues

[Phillips 1992] – I think convincingly – that the obvious errors are mostly not errors in

construction, but inaccurate repairs in ancient times or (more often) careless drawings or

reconstructions from fragmentary remains in modern times.


Not all of the composites are four-fold: Makthar (Kern 147) unites two bases in a semi-circle,

Gamzigrad (Kern 138) three in a hexagon, and Fribourg (Kern 136) eight serpentines within a

circle. But the principle of connecting the bases head-to-tail around the composite still obtains.

In all but two (at Makthar and in the House of Theseus at Kato Paphos), the path enters each

base from the innermost connecting course and leaves by the outermost connecting course –

which then jogs to the innermost course to enter the next base.

As another wrinkle, the base in the fourth quadrant is not always identical to the other three. In

the standard construction, the main axis (alone of the four) has two long connections along it: the

path from the exterior to the entrance of the first base, and the path leading from the exit of the

fourth base to the center of the composite (figure 11). A few of the mazes, however, avoid this by

rearranging the fourth quadrant so that the path enters and leaves that quadrant from the same

side (the inside of the composite), so that the second connecting path along the main axis is no

longer required (figure 15). The resulting pattern in the fourth quadrant does not correspond to

a SAT maze at all; it is in effect a maze with two separate serpentine paths to the center. Phillips

calls this the “Pompeian variation” after three examples from Pompeii (Kern 157–160). It is seen

not just at Pompeii – it appears also at Cremona (Kern 132) and Piadena (Kern 155).

Fig. 15: The Pompeian variation, after Kern 160

(Pompeii), with γ4

2 as the base

Fig. 16: The Pompeian variation, from a modern

T-shirt design, with γ4

1 as the base

The Pompeian variants are all early examples, from roughly the same time period. They are

conceivably products of the same artist, as the fudged fourth quadrant always follows the same

pattern (figure 17) – an ascending serpentine interwoven with a descending serpentine, closely

related to the Snake family (the η family): if the exit from the fudged quadrant were to lead

outward instead of into the center of the composite, the quadrant would be an ascending Snake.

This pattern is maintained even in a simpler modern design I have seen offered for sale on T-

shirts (figure 16).

Fig. 17: The Pompeian scheme in quadrant 4,

from (respectively)

the modern T-shirt design,

Kern 160 (Pompeii),

and Kern 157 (Casa del Labirinto, Pompeii)


Yet another variation: while most of the

labyrinths lead from the exterior to the

interior, two of the mosaics (Kern 138 from

Gamzigrad and 156 from Pompeii) are

represented by Daszewski and Kern as closed

loops. This variation may in fact be illusory.

Phillips regards the example from Pompeii as

an inaccurate early drawing based on a single

surviving quadrant, and the drawing of

Gamzigrad featured in Daszewski and Kern

does not agree with recent photos of the

labyrinth (figure 18), which display an open,

non-looping path.3

Fig. 18: The labyrinth at Gamzigrad

Let us look now at which SAT mazes appear in the compounds. Phillips was the first to offer a

close analysis of the bases. Daszewski had introduced a much simpler scheme, classifying bases

simply as serpentines (powers of γ2

), meanders (mostly powers of γ4

and γ6

), or spirals (mostly

single meanders larger than γ6

) 4 – and he failed to appreciate the Pompeiian variation as a clever

solution to the double axis problem, labeling such labyrinths merely as hybrids of meanders and

serpentines. Kern dismissed Daszewski’s scheme altogether, and evidently did not consider the

pattern of the bases to have any significance. As a result, I believe, he missed one of the informing

principles of the Roman design.

The main surprise in Phillips’s catalog is the overwhelming preponderance of the gammas and

their composites. Of the 47 surviving Roman-style composites, fully a third use the Classical-7

labyrinth γ4

2 as the base. Another third use other powers of γ


(i.e., other members of the Otfrid

sequence). A further sixth use odd serpentines (powers of γ2

), and in the final remaining sixth we

find a smattering of larger gammas or their composites (including the mixed Classical γ6


) – and

the lone case of α10

2. The larger Classicals are almost entirely absent: for the Romans the natural

route to expansion was repetition à la Otfrid.

From this we can state two reasonable conclusions. First, the Classical labyrinth was well known

to the Romans. The relative absence of simple Classical examples in the mosaics has led some

authors to surmise that it was not widely known – but on the contrary, its substantial presence as

a base in the compounds proclaims that it was widely regarded as the labyrinthine prototype. But

second, the preference for the Otfrid sequence over the larger Classicals makes me suspect that

Rome was not familiar with the extended seed patterns for the Classical sequence – probably

unlike Scandinavia, where Classical-11s handily outnumber the Classical-7s (and where Classical-

15 and Classical-19 are not unknown).

Finally, it is worth marveling a bit over the remarkable regularity of the Roman composite plan.

The principle of joining similar SAT bases head to tail around the center with two connecting

courses is reflected in the vast majority of the surviving mosaic labyrinths. Whoever first thought

of that really struck a sympathetic nerve in Roman culture.

Richard Myers Shelton, Roseville, MN, USA; September 2020



1. Authors have split on both terminology and visual metaphor for level charts: I draw my level charts

with the outermost course at the bottom of the stack and work upward, so that the number L is the

height of the stack. Phillips draws the outermost level at the top of the stack and proceeds downward

to what he calls the depth of the labyrinth.

2. The strong case for the modern provenance of the Domus Flavia “fountain” is made in [Lundén

2004]. The square Classical in the House of Fountains at Coimbra is described in [Lundén 1996,

30], and illustrated in [Saward 2003, 52]. The votive pillar from Smira (which also contains a round

labyrinthine pattern of interconnected swastikas) is described in [Shukriu 2010]. The simple

meander at Halstock, mentioned in [Phillips 1992], is described and illustrated in [Rainey 1987, 87,

plate 7]. The polychrome compound mosaic in Mérida, mentioned in [Saward 2003, 57], is in the

Casa del Anfiteatro, recently opened to public view, see [Ángeles Morcillo 2020]. The fragmentary

compound mosaic in Huete is described in [Torrecilla Aznar 2008]. The labyrinth at Bulla Regia is

mentioned in [Molholt 2011], footnotes 7, 16, and 66, with reference to [Hanoune 1993], which has

several diagrams. (My thanks to Jeff Saward for several of these references.)

My list does not include the well-known graffiti in Pompeii (Kern 107–108), nor several simple

Classicals of dubious Roman provenance or date, see [Lundén 1996], nor several labyrinthine

mosaics of interconnected swastikas (another popular mosaic pattern). Phillips limits his analysis

[Phillips 1992] to mosaic labyrinths in Chapter VI of Kern, so he does not include Side or the

Palatine fountain or Ravenna; he also omits Mieza and Salinas de Rosio, which were added to Kern

after his article appeared. Table 2 in Phillips contains a typo: the base for Daszewski 1 (Annaba)

should be γ2

4, not γ


2. In addition, the frequency counts for types γ


, γ4

2, and γ


3 in his Table 1 should

be 2, 15, and 7, respectively. (The total 43 is correct.)

3. The issue of closed loops deserves close attention, as it highlights a broader difficulty in interpreting

these often poorly-preserved labyrinths. The diagram from Pompeii records one of the first designs

found in the 1700s, and the original mosaic has long since vanished. Phillips argues that even if all

four axes in the original could be seen as single paths in what remained of the mosaic, this might

well have been another example of the Pompeiian variation, not a closed loop. Although meanders

are often presented as closed loops, there are no unambiguous examples among labyrinths, so there

is no strong evidence that Roman labyrinths ever contained closed loops. (In his discussion of

Gamzigrad, Phillips describes the carving at Side in Turkey as a closed loop, but high-resolution

photos show clearly that this is not the case.)

Kern attributes the drawing of Gamzigrad to R. Sobolewski. Kern gives no date for the drawing,

but the mosaic was discovered in 1953–1954. The site in Serbia is one of the best-preserved Roman

sites in Europe, though its importance was realized only after Daszewski and Kern were first

published: an inscription uncovered ca. 1985 identified it as Felix Romuliana, the palace-fortress of

Emperor Galerius (who, like several high Imperial officials of the time, was a native of Illyria in the

Balkans). Until his death in 311 CE, Galerius was the senior Augustus following the abdication of

Diocletian in 305. The palace was built about that time, yielding a solid date for the labyrinth. The

complex was designated a UNESCO World Heritage Site in 2007, and the reconstructed labyrinth

mosaic is currently on display at the site.

But Gamzigrad teaches us to be careful about jumping to conclusions. The drawing and the

reconstruction disagree about whether the path is a closed loop – and there are many other

differences, including something as basic as the spatial relationship between the hexagonal

labyrinth and the surrounding rectangular frame. Was the drawing done too hastily? Did it take

care to indicate the details and the places where the mosaic was not intact? Or conversely, was the

reconstruction too heavy handed, influenced by assumptions motivated by the need to have an

attractive finished product for display? Without a clear photograph of the mosaic as discovered, it

is impossible to give a definitive answer.


Similar questions arise with several of these labyrinths, where early drawings disagree with later

restorations. Some specimens have even been completely rebuilt as the original mosaic

deteriorates. In others (e.g., Kern 128 with base γ4


at Coimbra) it is possible to see the details of

reconstruction change over time, as early photos differ from later ones. So caution is advised!

4. While simple, Daszewski’s classification (p. 41–45) is not entirely consistent. In particular, the

dividing line between bases for meanders and bases for spirals is not clearly drawn. Daszewski

classifies two powers of γ6

as meanders (Syracuse/Taormina γ6


and Sarajevo/Stolac γ6


) and two

large single gammas as spirals (Al-Asnam γ10

and Dellys γ14

) – but then he includes Giannutri γ6


among the spirals, even while admitting that the base is not a single meander but divided into two

parts. It differs only marginally from γ6

2 – so why is it placed with the spirals and not the meanders?

Further, the exceptional Sousse α10


(probably perceived as having a base with four copies of γ4

) is

included among the meanders, though its essential difference from the others is thereby



Ángeles Morcillo, M. “Mérida culmina su triángulo arqueológico.” Hoy, 19 June 2020,

See especially photos 12 and 13 in the accompanying photo gallery.

Website of Monika Bugs. Katzenlabyrinth:

Daszewski, Wiktor A. Nea Paphos, Vol. II “La Mosaïque de Thésé.” Warsaw: Editions Scientifiques

de Pologne, 1977.

Hanoune, Roger. “Décor du monument: Les pavements mosaïques,” in Recherches archéologiques

franco-tunisiennes à Bulla Regia, vol. 2, pt. 1, Les thermes memmiens: Etude architectural et histoire

urbaine, by Henri Broise and Yvon Thébert, Ecole Française de Rome, 1993, p. 245-71. Available

online at

Kern, Hermann. Through the Labyrinth. tr. Abigail H. Clay, ed. Robert Ferré and Jeff Saward.

Munich, New York & London: Prestel, 2000, (References to Kern are to image numbers, or to

page numbers where no image is provided. The numbering in earlier editions differs slightly.)

Lindenmayr, Franz. Mensch und Höhle (website).

Austrian labyrinths:

Lundén, Staffan. “The Labyrinth in the Mediterranean.” Caerdroia 27 (1996), p. 28-54.

Lundén, Staffan. “The Palatine Labyrinth.” Caerdroia 34 (2004), p. 7-14.

Molholt, Rebecca. “Roman Labyrinth Mosaics and the Experience of Motion.” The Art Bulletin,

Vol. 93, No. 3 (September 2011), p. 287-303.

Phillips, Anthony. Through Mazes to Mathematics.

Phillips, Anthony. “The topology of Roman mosaic mazes.” Leonardo 23 (1992), p. 321-329. (Reprint:

Michele Emmer, ed., The Visual Mind: Art and Mathematics. MIT Press, 1993, p. 65-73.)

Rainey, Anne. Mosaics in Ancient Britain. Newton Abbot, England: David & Charles, 1973.

Saward, Jeff. Labyrinths & Mazes. New York: Lark Books, 2003.

Shelton, Richard Myers. “Greys Court: an invitation to symmetry.” Caerdroia 40 (2010), p. 21-35.

Shelton, Richard Myers. “Basic Labyrinth Math.” Caerdroia 48 (2019), p. 37-49.

Shukriu, Edi. “Two labyrinths and Dardanian, Greek and Roman relations by Dea Dardanica’s

monument.” 5ème colloque international sur l’Illyrie méridionale et l’Epire dans l’antiquité,

Grenoble, 8–11 octobre 2008, CRHIPA (Centre de recherche en histoire de l’Italie et des pays

alpins), Vol. II. Paris: De Boccard, 2010, p. 571-575.

Torrecilla Aznar, Ana. “El mosaico del laberinto de Huete (Cerro de Alvar Fáñez, Cuenca).”

Zephyrus, LXI, January–June 2008, p. 197-214.


In Memoriam: Wiktor A. Daszewski (1936–2021)

We recently received news that Wiktor Daszewski died on 17 January 2021 in Warsaw at the age

of 84. Daszewski was a pillar of the Polish archaeological community, an archaeologist of

international renown, especially for his work in Cyprus and Egypt, and a tireless proponent of the

preservation of archaeological material at sites and museums around the world.

Wiktor Andrzei Daszewski was born on 1 November 1936 in the village of Horodyslawice, a few

miles south-east of Lwów (modern Lviv, Ukraine). He studied in Warsaw, Kraków, and Perugia,

and received a Master's degree from Oxford and a Doctorate from Warsaw. He served as the

Director of the Centre for Mediterranean Archaeology at the University of Warsaw, and taught

at the universities of Warsaw and Trier. His field work centered on the Hellenistic sites of Nea

Paphos in Cyprus and Marina el-Alamein in Egypt (just west of Alexandria), rescuing the latter

from bulldozers to pursue an excavation that continued for 20 years. Not content with finding

new artefacts, Daszewski lent his prestige and effort to preserve the old, working as a coordinator

for UNESCO heritage sites and helping to establish and organize several archaeological

museums around the Mediterranean.

Labyrinth enthusiasts know Dr. Daszewski principally for his work at Nea Paphos. He was the

Director of the Polish archaeological excavations in Kato Pafos and joint author of the multiple

volume account of the findings. He was the sole author of the second volume (Nea Paphos: II. La

Mosaïque de Thésée, Editions Scientifiques de Pologne, Warsaw, 1977), which discusses the richly

decorated labyrinth in the House of Theseus at Kato Paphos, but which more significantly places

it in the milieu of the labyrinth as a Roman phenomenon. Despite the passage of over forty years,

this extended essay on the subject remains the definitive account of Roman labyrinths. It served

as the primary reference for Hermann Kern’s chapter on Roman labyrinths, and the lion’s share

of illustrations in Kern's chapter come from Daszewski’s larger plates. Daszewski’s detailed and

generously – indeed, exhaustively – illustrated catalogue of the surviving examples sits in an

honored spot on the shelves of many of us – and continues to be taken down and consulted with


Richard Myers Shelton

The labyrinth mosaic

preserved in the

so-called House of

Theseus, at Paphos,

Cyprus, excavated by

Wiktor Daszewski

in 1969


Jeff Saward, 2015


A Mysterious Medieval Maiden

Jill K. H. Geoffrion & Alain Pierre Louët

To find a woman’s face and neck in the centre of a

medieval manuscript labyrinth is most surprising!

Yet, on folio 80v of the thirteenth-century

manuscript known as Chantilly 0328 she is there,

shown from the side.

The woman in the centre of the labyrinth, Chantilly 0328,

fol.80v. Image courtesy of Musée Condé, Chantilly

Only two other historical labyrinths with women in

the centre have been identified. The first is a Roman

floor mosaic from the 3rd


century CE found in the

Paphos Archaeological Park on Cyprus. Ariadne has

been placed in the upper left watching the battle of

Theseus and the Minotaur (Kern 2000, 142 & 143).

Central panel of the mosaic from Paphos.

Photo courtesy of Cyprus Museum

The second is found in a fifteenth century fresco in a

church in Sibbo, Finland (Kern #601) where a

woman stands with half her body in the entrance to

the centre with her arms and head in the bottom half

of the eleven-circuit labyrinth’s centre.

Other visual elements surrounding the women in

these two labyrinths help with the interpretation of

their presence. Ariadne’s role in the labyrinth myth

of the battle between Theseus and the Minotaur is

well documented. Kern sheds light on the Sibbo

woman, noting that below the fresco “a

Jungfrudans,” a maiden’s dance, is shown. The

Cretan-type labyrinth has 11 circuits, and a small

female figure is depicted at the centre. She is clearly

the maiden around whom the dance is centred.”

(Kern 2000, p. 281)

The labyrinth fresco from Sibbo, Finland

The question of the meaning of this medieval depiction of a pretty woman’s head and naked

neck in the centre of the labyrinth with her rosy cheeks and orange hair set against a blue

background seems to hinge on the question, is she the personification of good or evil? To

use labyrinth symbology, is she more of a Theseus or Minotaur figure?


During medieval times, women were rarely held in the same regard as men and were often

considered to be agents of Satan (Jean Delumeau, La peur en Occident. Collection Pluriel.

Éditions Fayard, 1978, see chapter 10: Les agents de Satan III. – La femme, p. 398-449).

Using makeup was strictly forbidden in the Middle Ages and seen as the work of the Devil

because the human face was considered to be created in the image of God, thus altering its

appearance was thought to be disrespectful of the Creator. In the centre of this labyrinth,

we find a woman’s face with painted lips and extra rosy cheeks. Her head is encircled with

wavy reddish-orange hair that flows down her neck. The colour of her hair must be

understood through the lens of medieval colour symbology. Red was seen as the opposite

of white which represented all that was good and pure. Thus, red was linked to all that was

bad, including the Devil, demons, falsity, and betrayal. Yellow was also used to express

negativity, and when red and yellow were combined to make orange the negative meaning

of the colour was amplified. Orange was used to show the scandalous nature of a person and

was thus linked with prostitutes whose reddish-orange hair identified them as such, Judas

Iscariot, executioners, and all who were considered outsiders. (Michel Pastoureau, “Tous

les gauchers sont roux” in Le Genre Humain, 1988/1-2, No. 16-17, p. 343-354.)

The symbol of evil in the centre of the labyrinth has a long tradition, as we discussed in our

previous article “The Beast Within” (Caerdroia 44, 2015, see page 17 for a list of manuscript

labyrinths with only the Minotaur in the centre). From the ninth through thirteenth

centuries there are 12 manuscript labyrinths in which the Minotaur, symbol of danger and

evil, reigns alone in the centre. That we would find a twist on that theme in the thirteenth

century where the Minotaur has been replaced by the depiction of evil in the form of a

woman would be unique, and to modern sensibilities disturbing. The understanding and

depictions of the centre as a place of danger was later replaced in many manuscripts with

images that showed it as a place of victory over evil. As a universal symbol, it is not surprising

that the labyrinth’s meanings cannot be reduced to

a singular interpretation. (See: Geoffrion & Louët,

“Medieval Marvels: Fifty-Three Eleven-Circuit

Manuscript Labyrinths,” Caerdroia 49, 2020)

Who or what does this female represent? If we

look to context, we find a surprise unknown in

other medieval labyrinth pages. The text

surrounding this labyrinth, starting on the previous

folio and continuing onto the page with the

labyrinth is a recipe for a chicken pie. The link

between the woman and this particular recipe is far

from clear! The ambiguous identity and

significance of this female in the centre of the

labyrinth remains mysterious.

Folio 80v, Chantilly 0328.

Image courtesy of Musée Condé, Chantilly

Jill K. H. Geoffrion, Wayzata, MN, USA. Email: [emailprotected]

Alain Louët, Chartres, France; May 2021. Email: [emailprotected]




The Minnie’s Gap Labyrinths

Kirk Astroth

Just 1/10th of a mile north of the Utah state line with Wyoming near Flaming Gorge Reservoir

along US Highway 191 are two well-etched labyrinth images. Francois Gohier, a professional

photographer, first alerted me to the existence of these side-by-side images. They are located on

a sandstone cliff face, about 20 feet above ground level and above a large solitary boulder which

has a number of Indigenous rock images scattered about it, but none that resemble a labyrinth.

The labyrinth images are the only ones on the upper sandstone face.

Both these images appear to have been etched into the rock with metal tools; one is square while

the other is oval. What is also unique is that they both are dated – “1896” is etched into the centre

of the square version, and “96” is etched into the centre of the oval one. It is evident that a ‘seed

pattern’ utilizing dots was employed in the creation of the square image, but no such dots are

visible on the oval version. Both are being invaded by lichen growth suggesting that they have

been there for a time.

The Minnie’s Gap

labyrinth inscriptions,

Wyoming, USA.


Kirk Astroth, 2018

Although it is difficult to discern why these images were engraved with the date of 1896, it is

interesting that this is the year in which Utah finally became a state. More interestingly, perhaps,

is that this area in the late 1800’s was a hive of outlaw activity. Butch Cassidy and his gang

frequented the area since it is adjacent to Brown’s Park where they had several hideouts. Brown’s

Park is along the Green River and, with its cliffs and mountains, serves as a natural fortress from

the long reach of the law. Its lush valleys provided ample grazing pastures for rustled livestock,

even today, though, the area is remote and not easy to access.

Several miles to the west of Brown’s Park is Minnie’s Gap, named for Minnie Crouse Rasmussen,

the daughter of Charley and Mary Crouse who were early settlers in Brown’s Park. Minnie’s Gap

is located in a narrow notch between sandstone ridges and 7,000-foot peaks. It was once the

location of a small store that catered to the outlaws and anyone else traveling through the area.

Crouse Creek in the Brown’s Park area is also named for this prominent family. Charley Crouse

was a friend and associate of the Wild Bunch who is said to have aided the outlaws. At an alcove

in the cliffs called Cassidy Point, a cabin was built that was out of sight but commanded a sweeping

vista of Brown’s Park. A trail from the Crouse ranch led up to the alcove and it was Mrs. Crouse

who cooked meals for the outlaws hiding there and her daughter, Minnie, carried the food up the

trail to them.1


This area was a popular hang-out for a number of gangs at the time. It was here that Tom Horn

murdered Isom Dart, one of the few Black outlaws in the area.2 On August 18, 1896, Butch

Cassidy proposed to create a large gang out of the existing smaller gangs and to be called the

Train Robber’s Syndicate. Later it would become known simply as the Wild Bunch. During this

organization meeting, it is reported that over 200 outlaws were in attendance. The Hole-in-the-

Wall Gang was led by Flatnose George Curry and included Kid Curry (his son) and the Sundance

Kid. Also in attendance were members of the Powder Springs Gang, the Blue Mountain Gang,

the Robber’s Roost Gang and the Diamond Mountain Boys which was Butch’s gang at the time.

The Crouse Ranch was transformed into an armed camp.3 However, a dispute about who should

lead the gang resulted in everyone dispersing to go on a rampage to prove who could rob the most

banks and trains. They reconvened August 18, 1897 to see who was most successful who would

then become the leader of the Wild Bunch.4

Minnie Crouse was a girlfriend of Butch Cassidy’s and apparently ran a small store that catered

to the outlaws who hung out in nearby Browns Park.5 Minnie lived in the area for a number of

years and provided a colourful oral history in 1978 of her times and experiences in the Brown’s

Park area.6,7

Why two perfectly engraved labyrinth images are found here is still a mystery. They were unlikely

to be etched into the rock by any of the outlaws who were too busy evading the law and planning

bank and train robberies. Perhaps someone did this to mark Utah’s entry into the Union. Or

perhaps some local was familiar with William Lethaby’s book, Architecture Mysticism and Myth

published in 1892, where two labyrinths, one square, one circular, were illustrated. Perhaps these

two images were inscribed much later and simply done to commemorate Utah’s admission as a

state. Mormonism itself is heavy with Masonic imagery and ritual, and the labyrinth image is a

common motif in Masonry, symbolic of progress, travel, and movement.8

Kirk Astroth, Tucson AZ, USA; April 2021

Email: [emailprotected]

Two labyrinths (coins from Knossos), one square,

one circular, as illustrated in William Lethaby’s

Architecture Mysticism and Myth, published 1892


1. Redford, Robert. The Outlaw Trail. New York: Grosset & Dunlap Publishers, 1978, p. 129-130.

2. Ibid.

3. Ibid.

4. Ibid. p. 131.

5. Ibid. p. 130.

6. National Park Service. John Jarvie of Brown’s Park, Chapter Three (2008). Available online at:

7. Utah Division of State History, The Uintah County (Utah) Oral History Collection, 1974-2002.

(2005). Available at:

8. Zeldis, Leon. Masonic Symbols and Signposts. Lancaster, VA: Anchor Comms., 2003, p. 82-90.


Notes & Queries

Our regular round up of matters labyrinthine brings together short contributions and notes from

Caerdroia readers worldwide, also items from the Labyrinthos Archives that require further

research, or simply deserve recording. Similar notes and queries are welcomed for future editions.

A Labyrinth Inscribed Powder Horn Jeff Saward

Formerly in the collection of Rich Nardi (, an

inscribed powder horn, a little over 13 inches (33 cm) long and dating from the American

Revolutionary War, recently sold at auction for $5,760. The horn was originally the property of

Henry Thorn, and his name and the date 1779 are engraved along one side and a fort, sailing ship

and soldiers holding a sword, raising a flag and aiming a flintlock musket at a tree appear around

the body of the horn. Of particular interest is the classical-type labyrinth is engraved at the widest

end beneath a scalloped border. The fine detail preserved by the engraving process clearly shows

that much of the design of the labyrinth was executed with a compass, and a series of small dots

running through the upper half provide some clues to how Henry

Thorn engraved the design. It is known that he was born in

Rhode Island in 1759, enlisted at the age of 17, and served with

the troops that wintered with George Washington at Valley

Forge in 1777-78, before being discharged in New Jersey in 1780.

This fascinating object provides one of the earliest dated

occurrences of the labyrinth symbol on the American east coast.

Henry Thorn’s Horn, dated 1779. Photos courtesy of Morphy Auctions

A Labyrinth in Myanmar a note from Klaus Aarsleff, Svinninge, Denmark

The Dhamma Yazika pagoda in the old royal town of Bagan

in Myanmar (Burma), was constructed as a solid brick

building in 1196 CE. There are, however, some small

chambers surrounding the building, and on the south side a

life size statue is found in one of the chambers. On the wall

adjacent a perfect classical labyrinth has been painted, and

script in ancient Burmese next to it reads (according to my

guide): “The entrance is here, to the South.” As there is no

entrance, as the pagoda is solid, this must be interpreted

symbolically, and it’s interesting to note that several more

labyrinths have been scratched alongside as later graffiti.


Another Labyrinth on an English Sampler photo from Kathy Andrews

A third sampler decorated with the same labyrinth design (see

“Two Labyrinths on English Needlework Samplers,” Caerdroia

43, p.4-6), and again produced in Dewsbury, Yorkshire, has

recently been brought to our attention. In a private collection in

Iowa, USA, this small sampler was created by Mary Parsons “In

The Year Of Our Lord 1780 in Dewsbury Yorkshire” and along

with the usual alphabet and numbers (one to fifteen) is decorated

with a typical morally instructive verse. The lower panel of the

sample is filled by a square labyrinth and a depiction of Noah

standing on the deck of his ark, complete with dove and olive

branch. Both the ark and the labyrinth are identical in form to

those on the two previously documented samplers, and

interestingly the labyrinth contains the same distinctive quirk – a

slight offset of one turn on the right-hand side, not present on any

of the engravings in contemporary books. This surely suggests

that Mary Parsons was using the same stitching pattern as Mary

Blackburn in 1785 and Ann Lewis in 1798, and that maybe she

was also a pupil of Mrs Lees?

With three samplers decorated with identical labyrinths now on record, stitched over a period of

nearly 20 years, it seems likely that more might yet be recorded, although to the best of our

knowledge, none have ever been documented from anywhere other than Dewsbury. Might this

suggest that this specific design was a speciality kept in Mrs Lees’ library for her girls, rather than

a template in a more widely available work?

The Modern Labyrinth Jeff Saward

Undoubtedly, many readers of Caerdroia will

already have an opinion as to what constitutes a

‘modern labyrinth,’ but 120 years ago a remarkable

fairground maze of that very name was installed at

Vlissingen (Flushing) in the Netherlands, a popular

tourist destination, with day trips by steamer from

Sheerness in Kent, England. The Modern Labyrinth,

probably a wooden panel maze, but maybe formed

of mirrors, complete with an ornate art deco façade

and a tall central tower surrounded by a spiral slide

called the Toboggan, was operated by G. de Klerk-

Hoevens and set up at the Zeilmarkt (sailing

market) around 1900. It was still advertised in a local

newspaper (Vlissingsche Courant) in July 1905, so

may have been in existence for several seasons. Two

photographs exist in the Zeeuws Archief

photographic archive (collection no. 23501 & 41288)

and my thanks go to Marius Voet for spotting these

and helping with research and translation.

The Modern Labyrinth, Vlissingen, ca. 1900.

Photo courtesy of Fotocollectie Vlissingen, nr 41288


Submissions to Caerdroia

Caerdroia is always pleased to receive material for publication. Readers are urged to submit

papers, shorter articles, notes, information, photographs – indeed, anything labyrinthine –

for possible publication in future editions of Caerdroia. Articles and notes should preferably

be sent as e-mail attachments in Microsoft Word .doc or .docx format (although .rtf and

similar formats are acceptable).

Illustrations and photographs are preferred in .jpg or .tif format at 300 dpi resolution please,

but please keep illustrations separate from text, and send as separate files, with position in

text clearly marked. Photographs: colour or b&w prints and 35mm transparencies are also

welcome if digital versions are unavailable. A preferred style guide for authors is available

on the Caerdroia Submissions page on our website:

Because Caerdroia is a specialised journal for enthusiasts, no payment can be made for

submissions, but any reproduction fees required will be covered, and all significant

contributors will receive a complimentary copy and/or digital PDF. Short notes and press

clippings are likewise welcomed, along with plans, postcards, guide books, photographs, etc.,

from any maze or labyrinth you may visit, for addition to the archives. Deadline for inclusion

in Caerdroia 51: December 2021 please, for scheduled publication Spring 2022.

Subscription to Caerdroia

As an enthusiast’s journal dealing with a specialised subject, Caerdroia relies on reader

subscriptions to allow it to continue to provide a forum for maze and labyrinth research and

news. Subscription provides the next edition of Caerdroia (and access to the online digital

version) and supports the production of the journal, maintenance of the Caerdroia Archives,

covering all aspects of mazes & labyrinths worldwide, and our extensive website. A scanned

reprint service from out-of-print editions is also available to subscribers. The annual fee is:

£8 UK, €10 Europe, $15 USA, £12/US$15 (or equivalent) rest of world.

• Payment details: UK: all cheques payable to LABYRINTHOS please, or subscribe online

via Paypal – visit our website: – payment via

electronic transfer also welcome, please contact us for bank details

• Europe: payment in Euros is very welcome (fold in dark paper for security), by electronic

bank transfer or subscribe online via Paypal – visit our website or contact us for details.

• USA: online subscription via Paypal is preferred – visit our website or contact us for details.

US$ bills are also accepted, but please fold in dark paper for security.

• World: online subscription via Paypal is preferred – visit our website or contact us for

details. Readers in Canada, South Africa, Australia, N.Z. & the Far East are welcome to

pay in UK£ or US$, but please fold in dark paper for security.

Readers in Scandinavia may subscribe by applying to our agent John Kraft:

Rolighedsvej 19/1th, 1958 Fredriksberg C, Denmark. Email: [emailprotected]

Subscription or payment queries? - contact us at: [emailprotected]

Caerdroia on the Internet:


53 Thundersley Grove, Thundersley,

Essex SS7 3EB, England, UK.

Telephone : +44 (0)1268 751915

E-mail : [emailprotected]

Website :

The story of mazes and labyrinths is as long and tortuous as their plans might suggest.

For many, mention of the labyrinth may recall the legend of Theseus & the Minotaur.

An increasing number will know of the ancient labyrinth symbol which occurs around

the world, at different points in time, in places as diverse as Brazil, Arizona, Iceland,

across Europe, in Africa, India and Sumatra. This symbol and its family of derivatives

have been traced back 4000 years or more, but its origins remain mysterious. Modern

puzzle mazes, however complex their form, are but the latest episode in this

labyrinthine story.

Labyrinthos is the resource centre for the study of mazes and labyrinths, with an

extensive photographic & illustration library and archive, offering professional

consultation and services for owners, designers, writers and publishers and consultation

for labyrinth design and installation. Contact Jeff Saward or Kimberly Lowelle Saward

at the address above, or visit our extensive website for further

details of Labyrinthos and Caerdroia.

Our annual journal Caerdroia, first published in 1980, is dedicated to maze and

labyrinth research and documentation. Produced by labyrinth enthusiasts for fellow

enthusiasts, it keeps in regular contact with correspondents throughout the world,

exchanging information and ideas, to help create a clearer picture of the origins and

distribution of the enigmatic labyrinth symbol and its descendants, from the earliest

rock carvings and artefacts through to modern puzzle mazes of ever-increasing

complexity and ingenuity.

Current subscribers to Caerdroia include maze and labyrinth researchers and

enthusiasts, archaeologists and historians, artists and authors, designers and owners,

and members of The Labyrinth Society. As a non-profit making journal dealing with a

very specialised subject, Caerdroia relies on reader contributions, submissions and

subscriptions for support. If you are interested in the history, development, diversity or

potential of mazes and labyrinths in any of their forms, perhaps you would care to join

us on the path....

Jeff Saward & Kimberly Lowelle Saward, Labyrinthos


Caerdroia is an independent

journal for the study of

mazes & labyrinths

Established 1980

Published annually

Produced by & ©

Labyrinthos 2021

CAERDROIA 50 - - [PDF Document] (2024)
Top Articles
Latest Posts
Article information

Author: Lilliana Bartoletti

Last Updated:

Views: 6089

Rating: 4.2 / 5 (73 voted)

Reviews: 80% of readers found this page helpful

Author information

Name: Lilliana Bartoletti

Birthday: 1999-11-18

Address: 58866 Tricia Spurs, North Melvinberg, HI 91346-3774

Phone: +50616620367928

Job: Real-Estate Liaison

Hobby: Graffiti, Astronomy, Handball, Magic, Origami, Fashion, Foreign language learning

Introduction: My name is Lilliana Bartoletti, I am a adventurous, pleasant, shiny, beautiful, handsome, zealous, tasty person who loves writing and wants to share my knowledge and understanding with you.