Roman silk block damasks

Roman silk block damasks1
Susan J Foulkes, Durham and Online Guild
When I am weaving I feel connected to the past in a tangible way. Weaving tools,
looms and cloth have their own history and, as I have discovered, so do weave
structures. I love to weave scarves and my favourite structure is a block 3/1 - 1/3 twill.
In one block, warp floats are created on the surface of the cloth. This is contrasted with
the other block where weft floats predominate. The patterns made by the blocks are
emphasised by using different colours for warp and weft and the effects of light
This weave structure is an example of damask and seems to have originated within the
Roman Empire. It is likely that the Latin textile term scutulatus (diamond or lozengeshaped, chequered) refers to these simple check patterns (Wild, 1964). Weaving
damasks was a skilled job. The entry in the Edict of Diocletian (301 CE2) gives the
wages of a woman worker in silk scutulata as 30 – 50 denarii per day with maintenance.
In comparison, an elementary teacher only received 50 denarii per month for one pupil!
(Tenney, 1940).
Our knowledge of the early silk weaving which used this weave structure is, quite
literally, fragmentary. Fifteen such fragments, ranging from the second to the fourth
century CE have been found in Palmyra (Syria), England, Germany, Switzerland,
France and Hungary (De Jonghe, 2001). Intrigued by the possibilities presented by these
archaeological remains, I decided to weave samples of the six silks which have a pattern
published in the academic literature (see table). I wanted to bring the black and white
diagrams to life.
Palmyra (Syria): S 38
Holborough, Kent
Spitalfields, London
Conthey, Switzerland
Trier, Germany (purple)
First fragment
Trier, Germany
(yellow) Second
Approximate date
Number of
heddle rods
3rd century CE
250 CE
3rd – 4th century
4th century CE
No later than 395
No later than 395
Warp ends
per cm
Picks per cm
Patten repeat
50 – 60
Not known
50 - 60
Not known
Not known
54 picks
Not known
50 - 60
66 picks
128 picks
96 picks
This is a more detailed article than the one which appears in the printed Journal and it is only available
from the Journal website
CE (Common Era) and BCE (Before Common Era) are equivalent to AD/BC and are commonly used by
historians in academic publications.
Journal for Weavers, Spinners and Dyers 233, Spring 2010
For two-block patterns eight heddle rods are required. The Romans used a vertical
loom (similar to a modern tapestry loom) which had a warp and cloth beam. Silks may
have been woven on such a loom, although it is difficult to imagine that 20 heddle rods
could be accommodated or used. Tela holosericis vestis scutulatae („a loom for
weaving pure silk damask cloth‟) was the most expensive loom described in the copy of
Diocletian‟s Edict found in Phrygia, part of modern Turkey (Wild, 1987). This is
assumed to be a type of horizontal loom with warp and cloth beams, without a reed, but
with shed sticks and simple heddle rods which would have rested on the warp when not
needed. More than one person must have been required for the weaving process – the
weaver manipulating the weft and one or more assistants lifting the required heddle rods
and, perhaps, then inserting a series of pattern sticks in advance. This could enable the
weaver (or another assistant) to lift the correct shed for each pick.
Whatever type of loom was used, the weaver had to make heddles for the individual
warp threads. Heddles are made when the warp is under tension. Nowadays, handloom
weavers make a singles cross for all warps regardless of the complexity of the
threading. In the absence of a loom with a super-structure of eight shafts, a different
method of warping is required. A direct way of separating the warp threads into blocks
is to wind the warp over the requisite number of posts which correspond to the number
of heddle rods. Then the warp ends on each post are transferred to the appropriate
heddle rod so that, once the warp is under tension, the heddles for the individual threads
can be made. The pattern, therefore, is made during the warping. If a singles cross is
made, the task of separating the warp threads onto the correct heddle rods in each block
would have been extremely difficult and error prone. On examining the published
patterns of these ancient silks, there are relatively few mistakes in the warp. More
mistakes occur in the weft.
The modern handloom weaver uses warp threads in the blocks and weft picks in
multiples of four. This means that the block outlines are always cut cleanly. However,
the earliest known fragments of 3/1 block twills do not have this characteristic. The
understanding of the essential structure of 3/1 block twill weaves appears to have
developed over time. All of the early fragments have warp blocks which are multiples
of four ends, but the number of weft picks per block varies.
When we examine the weft in these fragments, an unusual feature is that the number of
picks per block varies; the more usual multiples of 4 were complicated by blocks of 6,
Journal for Weavers, Spinners and Dyers 233, Spring 2010
9, 11, 14, 15, 17, and 26 picks. This must have made the weaving more difficult since
the order for lifting the heddle rods would vary. As another complication, the direction
of twill, Z or S, could also change. In the purple silk from Trier, for example, both warp
and weft faced blocks could have either S or Z twill directions depending upon their
place in the overall pattern. With an 128 pattern repeat, it must have been a difficult task
to plan and weave this damask pattern without a computer programme and a modern
loom! Making the warp and lifting the correct sequence of heddle rods would have
been highly skilled jobs.
Having researched six silks in detail, I wove the patterns on a 32 shaft Megado
electronic dobby loom. These are not exact reproductions, as the spun silk I chose is a
thicker yarn than in the Roman examples, and some patterns needed to be adapted. For
warp and weft I used 2/60 silk (used double), sett at 36 epi (14 epc). This makes the
check patterns larger than in the originals.
Palmyra (Syria)
The earliest silk in my group is fragment S38 from Palmyra, dated to the 3rd century CE
(Pfister, 1937). In this fragment the silk has no twist in either warp or weft, which is
unusual. The warp blocks are in multiples of four warp threads (eight threads in each
block), but the weft sequence is 15, 9, and 9 picks. The edges of each block cut cleanly
(the binding points are in opposition) but the fragment is so small that the full pattern
repeat is unknown. To replicate this textile pattern, I added 9 picks (indicated by the
brown weft thread) to make a regular pattern repeat (see diagram A).
Diagram A: Kent.
Silk from Palmyra
Woven sample of
Palmyra silk
Holborough, Kent
Journal for Weavers, Spinners and Dyers 233, Spring 2010
The second silk is the fragment from Holborough, Kent, and is also dated to the 3rd
century. The weave structure is a broken 3/1 twill. The fragment is small with no
selvedges so it is assumed that the lightly spun pale brown yarn is the warp and the
unspun darker brown yarn is the weft. The warp threads are in blocks of 8 and 16 and
the weft picks are in blocks of 6, 6, 20, 6, 8, and 8 (Wild, 1965). To modern weavers
the sequence of weft picks is unusual and creates difficulties in producing a cleanly cut
block outline. The Holborough fragment has one uneven edge to the blocks which
shows on picks 12 - 13 (see diagram B). However, the fragment is too small to be sure
whether this is a weaving fault. This unevenness can be easily corrected by a different
sequence of lifts for the second block of 6 picks (see diagram C). This slight „fault‟
might not have been noticeable with such a high warp and weft count.
Diagram B: The Holborough Silk
Diagram C: Corrected Holborough
Silk pattern
Woven sample of Holborough
Journal for Weavers, Spinners and Dyers 233, Spring 2010
Spitalfields, London
A related sample is that found in 1999 at Spitalfields, London. Sometime in the 3rd – 4th
century, a young Roman girl in her early 20s had been buried in a lead coffin. This is
now beautifully displayed in the Museum of London. With her was found a small piece
of silk damask, measuring 2 cm x 5 cm, which was probably part of a head band. It has
a similar weave pattern to the Holborough silk including the same „fault‟ line (Wild,
2009) which appears at picks 41 and 42 (see diagram D). The weft sequence is in blocks
of 17, 24 and 14 picks. The diagram shows the sections of pale blue warp and lilac weft
(14 picks) which I added to make a pattern repeat.
Diagram D: Spitalfields silk
The same„fault‟ line as on the
Holborough silk, appears at picks 41
and 42.
Woven sample of Spitalfields Silk
Conthey (Switzerland)
The fourth example is that from Conthey in Switzerland. I have altered the pattern
slightly as the picks for the pattern fragment are 11, 12, 12, 26, 14, & 12 (diagram E). It
is easier to weave the pattern when the picks are in multiples of 4.
Journal for Weavers, Spinners and Dyers 233, Spring 2010
Diagram E: Conthey silk
Woven sample of Conthey silk.
Trier, Germany
Examples five and six are fragments from the tomb of St Paulinus. He was made
bishop of Trier in 349 but he was exiled by Emperor Constantius II to Phrygia where he
died in 355. His remains were returned to Trier, where they were reburied in 395, with
two silk coverings. These were examined in the 19th century. The first sample from
Trier is a purple silk in a more complex pattern than all the other fragments (De Jonghe
& Tavernier, 1978). It requires 20 shafts, but contains a number of weaving faults which
I eliminated by adapting the pattern. This ambitious pattern is not entirely successful.
Once woven, the shifting twill direction seems to work against the block design and
makes the overall effect less dramatic.
Diagram F: First Trier fragment
showing weaving errors
Journal for Weavers, Spinners and Dyers 233, Spring 2010
Woven sample of first Trier silk.
Diagram G: First Trier silk
It is reported that the Roman Emperor Aurelian (270 - 275) would not have any
garments made wholly of silk nor give them as presents. „When his wife besought him
to keep a single robe of purple silk, he replied, “God forbid that a fabric should be worth
its weight in gold.” For at that time, a pound of silk was worth a pound of gold‟ (Magie,
1968). The Edict of Diocletian, an early unsuccessful attempt to control inflation by
fixing the maximum prices for goods and services, lists the cost of white silk thread at
12,000 denarii for one pound but purple silk thread cost 150,000 denarii per pound.
Added to this, spinning purple silk (the Romans always added a twist to the warp silk)
cost 116 denarii for one ounce. This purple silk buried with St Paulinus was an
extremely costly and exclusive item.
The second fragment from Trier is in yellow silk (Wild, 1987). This pattern requires 16
shafts and the blocks are made of 8 warp threads and 8 weft picks like modern versions
of this weave. I first saw an old black and white photograph of this fragment (Wild,
1987 p. 463) and attempted to work out the pattern. I wove a sample and, on a visit to
Trier, managed to track down the original. The Dom Museum staff were wonderfully
helpful and the silk was brought out of storage by the professor himself. Along with
other fragments of silk found in the tomb it had been mounted during the nineteenth
century and was placed between two sheets of glass in a large ornate picture frame.
Unfortunately, the pattern on the silk has now faded to invisibility. However, the frame
Journal for Weavers, Spinners and Dyers 233, Spring 2010
contained a graphed diagram of the pattern which, to my delight, seemed to correspond
to the pattern that I had produced.
Diagram H: Second Trier fragment
Sample of the second Trier silk on the
I have used this pattern many times and it produces a stunning effect when woven with
one colour. The pattern appears to dissolve into circles, squares and crosses. The effect
is most pronounced when the block size is small i.e. 8 threads and 8 picks and using a
fine thread. When a larger block size is used, the optical effect is altered. Using 16
shafts for the pattern and 2 shafts for a selvedge, I wove a purple and magenta fabric of
2/60 silk (used double), sett at 40 ends per inch. This was made into the chasuble
illustrated below.
Chasuble for St Paul‟s Church,
Hunwick, County Durham
Detail of chasuble
Journal for Weavers, Spinners and Dyers 233, Spring 2010
Researching Roman silks provided me with more detailed information about the
fragments from Trier. I found that I had simplified the pattern for this second fragment:
the original had 12 picks for each block, not 8 as I had assumed. This silk pattern is the
only one of the six fragments that conforms to our modern use of 3/1 block twill with
both warp and weft ends being in multiples of 4.
It has been argued that the fragments found in England, Switzerland and Germany are
the products of one weaving centre because they show similar characteristics and
development of the weaving technique. One theory is that they all originated in the
imperial gynaecium (workshop) in Trier (De Jonghe & Tavernier, 1978).
In weaving these silks, I have been fascinated by the links between past and present:
learning that my favourite weave structure originated within the Roman Empire and
discovering that the Romans had a more complex loom than is generally realised. But
most importantly, through the cloths themselves – by weaving samples and seeing the
designs shimmer in the light – the past can be brought into the present. I hope that by
bringing them to a wider audience of weavers that the designs can be used once more
and their beauty appreciated.
De Jonghe, D. (2001) From the Roman horizontal loom to the 3/1 twill damask loom of
the early medieval period. In: The Roman Textile Industry and its Influence, eds. Walton
Rogers, P Bender
De Jonghe, D. & Tavernier, M. (1978) Les damasse de la proche-antiquite, Bulletin de
Liaison du CIETA, Vol 47, no 1, pp 14 – 42
Frank, Tenney. (1940) Vol 5 An Economic Survey of Ancient Rome, John Hopkins
Magie, D (translator) (1968) Scriptores Historiae Augustae volume III, XLV, Loeb
Classical History.
Pfister, R (1937) Nouveaux Textiles de Palmyre, Les Éditions D‟Art et D‟ Histoire:Paris
Wild, J, P. (1964) The Textile Term Scutulatus, Classical Quarterly, Vol 14, no 2
(Nov), pp 263-266.
Wild, J. P. (1965) A Roman silk damask from Kent, Archaeologia Cantiana, Vol 80,
pp 246 – 250.
Wild, J. P. (1987) The Roman Horizontal Loom, American Journal of Archaeology, Vol
91, no 3 (July), pp 459 – 471.
Wild, J. P. (2009) personal communication.
Note: The PCW weave programme was used to create the weave patterns.
Journal for Weavers, Spinners and Dyers 233, Spring 2010