You are viewing hrj

Previous Entry | Next Entry

doll
I got an interesting mini research question in e-mail this week, via my article Another Look at St. Louis' Shirt. The questioner asked, "Is there any way St. Louis' height could be approximated [from the garment], or does the design of the garment give nothing away?" And, as usual, I got sucked into the joy of rolling around in the data for a while and ended up with some interesting analysis but very few clear conclusions. The following includes a whole bunch of data and discussions that I hadn't included in my original answer (including more comparative material and the differences between Burnham's estimate of the garment's dimensions and my own estimates).

ETA: I've added a little more data regarding the relationship of sleeve length to shoulder width.

So here's a somewhat stream-of-consciousness presentation of how I sit down to analyze a question of this type.

The basic question is this: what characteristics of a garment provide reliable information about the size (especially, the height) of the person it was made to fit? There are two major aspects to this: What are the critical measurements of the garment? Given the style of the garment, how are those measurements intended to relate to the human body? A third, somewhat less important, question is: What bodily dimensions are most stable and useful for comparisons?

Let's start with the "typical" human body, as exemplified by DaVinci's "man in a circle" model and similar artists' proportions guidelines. This provides us with a template where height is approximately equal to arm-span and where the total height is approximately 7 head heights. Modifying this slightly to make it more relevant to clothing, garment length from the shoulder to the ground is approximately equal to the distance from wrist to wrist and is approximately 6/7 of the total height of the wearer. "Approximately."

Now let's look at the intended relationship of some typical garments to their wearers. During the general era of this garment (13th c. plus/minus), the sleeves of most body garments were full length, i.e., going all the way to the wrist. So the span from wrist to wrist of a garment should, at a minimum, be the span from wrist to wrist of the wearer. But it might be more. Some styles of sleeves were over-long and ruched up on the arms. Some styles of garments were very loose in the body and might be broader across the torso than that of the wearer. And in no case for this general style of garment is the garment likely to be "skin tight" across the torso. So the only real generalization possible is that it's unlikely that the garment "arm span" measurement will be smaller than the wearer's wrist-to-wrist measurement.

The vertical length of garments is more driven by fashion than the arm-span. For men, hems might be anywhere from knee-length to floor-length. Ecclesiastical garments (or ceremonial garments such as the regalia of the Holy Roman Emperors) tended to be full length, while everyday garments, especially for working purposes, tended to be shorter to varying degrees. So let's look at some possibilities:

A) full-length garment, sleeves/torso not over-large = armspan:height ratio is ca. 1
B) short garment, sleeves/torso not over-large = armspan:height ratio > 1
C) full-length garment, sleeves/torso over-large = armspan:height ratio > 1
D) short garment, sleeves/torso over-large = armspan:height ratio significantly > 1

So how do you tell the difference between (B) and (C)? Another tell-tale measurement is that from the top of the garment to the natural waistline. If the garment is relatively shapeless -- i.e., the body is either cylindrical or it flares evenly from top to bottom (i.e., like a cone) -- then you have no clue on this score. But most garments of the 12-14th century or so have at least some indication of the waistline based on the insertion of gussets into the skirts or by the cut of the main fabric. Jumping ahead to the results of my data-crunching, for an ankle-length garment, the ratio of this upper back length to the full garment length runs around 0.35 or so. So a larger ratio indicates a shorter overall garment. (A smaller ratio might indicate a "trailing hem" length of garment or it might indicate that the apparent "waistline" on the garment is higher than the natural waistline. For example, sometimes gussets start higher than the natural waistline. This can sometimes be identified based on the relationship of the top of the gusset to the sleeves.)

So lets look at some other men's garments from around the period of the St. Louis tunic. The Louis garment has some of its closest stylistic similarities to ecclesiastical albs, so I included several albs for which I had approximate measurements. As a secular comparison, I included two Danish outer tunics date to the 11-12th century. The garments are:

B = alb associated with St. Bernulf, held in the Netherlands, 11th c.
T = alb associated with St. Tomas Becket, held at Sens, France, 12th c.
I1 = Italian alb from a group dating to the 13-15th c.
I2 = Italian alb from a group dating to the 13-15th c.
I3 = Italian alb from a group dating to the 13-15th c.
I4 = Italian alb from a group dating to the 13-15th c.
I5 = Italian alb from a group dating to the 13-15th c.
S = Skjoldehamn tunic, Norway 10-11th c.
K = Kragelund tunic, Denmark, 11-12th c.
M = Moselund tunic, Denmark, 11-12th c.
Bk = Boksten tunic, Sweden 13-14th c.

The table below also has the estimated measurements for the Louis garment:
L (hrj) = my estimates
L (B) = Burnham's estimates

I converted all the measurements to inches (sorry metric folks!) and many of them are guesstimates but should be good enough for rough proportions. The following are the basic measurements and proportions. The measurements and proportions are as follows:

- Span = 2x sleeve length + width of garment at shoulder
- Height = shoulder to hem
- Upper back = shoulder to start of gussets
- Shoulder = width of garment at shoulder
- Wearer height = estimated height of the body found with the garment (if relevant)
- Span:height = ratio of arm-span to garment height
- Back:height = ratio of upper back length to garment height
- Length Fraction = ratio of garment length to wearer's height at shoulder (i.e., 6/7 total wearer's height); this gives an estimate of where the hem falls, e.g., a ratio of 0.9 is just above the ankle, 0.8 is around mid-calf, 0.7 is right around the knee

Item

Span

Height

Upper
Back

Shoulder

Wearer
Height

Span:Height

Back:Height

Length
Fraction

L (hrj)

46

46

16

16

*

1.00

0.35

*

L (B)

66

43

22

22

*

1.53

0.51

*

B

69

70

25

33

*

0.99

0.36

*

T

70

60

23

40

*

1.17

0.38

*

I1

88

66

*(1)

46

*

1.33

*

*

I2

69

70

*(2)

*

*

0.99

*

*

I3

89

76

33

40

*

1.17

0.43

*

I4

84

66

*(1)

28

*

1.27

*

*

I5

93

85

53

42

*

1.09

0.62

*

K

70

46

24

22

*

1.52

0.52

*

M

65

50

18

20

70

1.30

0.36

0.83

Bk

67

46

22

19

69

1.46

0.48

0.78

S

65

43

15

27

62

1.51

0.35

0.81



* measurement not available even by estimate
(1) no "waist"; (2) no figure available;

We're going to ignore the Louis garment for the moment.

Starting just with the span:height ratio, we have four garments around 1.00+/-0.10 -- the Bernulf and Italian albs #2 & #5. The rest fall in roughly three groupings of "excess sleeve length":

"slightly long" - ratio ca. 1.17 (Thomas alb, Italian #3)
"medium long" - ratio ca. 1.30 (Italian #1, #4, Moselund)
" really long" - ratio > 1.40 (Kragelund, Boksten, Skjoldehamn)

Keep in mind that "really long" can simply mean "short garment with full-length sleeves" which is the more likely explanation for the secular tunics. For the three secular garments where we have the wearer's body, the garment's percentage of "full length" runs around 80%, i.e., around mid-calf. (The Kragelund tunic would be similar if worn by a man the same height as the Moselund and Boksten men.)

Looking at the overall garment height, the majority of the albs run from 60-70" which, using the "body = 7 heads" estimate, corresponds to a total height of approximately 5'10" - 6'10". Two of the albs are longer. While there's a very remote possibility that the 76" one might conceivably have been made for an extremely tall man (over 7 foot) this is an implausible explanation for the 85" one and I think we have to add into our interpretations that this style of garment may be deliberately over-long (and bloused up over a belt or some similar style). This, of course, casts doubt on height estimates based on the "shorter" full-length albs. But the ecclesiastical style can be assumed to be full length (based on pictorial evidence) and the minimum 5'10" estimate for this group corresponds to the measured/estimated heights of the bodies found with the Moselund and Boksten garments. So as a "reality check" for a typical man of the day, this height might be a useful reference. If we use this reference height for the two albs suspected of being extra-long, then calculate the ratio of the span to an estimated shoulder-to-floor distance, then Italian albs #3 and #5 both belong in our "really long sleeve" category (with ratios of 1.48 and 1.55 respectively).

Looking at the total height of the secular garments, a reality-check suggests that these are not full-length garments. In the case of the Moselund tunic, we actually know that the body it was found on was 5' 10" tall, that is, if it had been full-length we'd have expected the garment length to be more like 60" rather than 50", in which case the span:height ratio would be 1.08 and so smack dab in our "normal" range. The same is the case for the 5'2" Skjoldehamn man: the span:height ratio for a full-length version of his tunic would be 1.05 and solidly "normal". Similarly, the Boksten tunic, if full length, would have a span:height ratio of 1.14 placing it between our "normal" group and our "slightly long sleeves" group. The Kragelund tunic is the same length as the Bosten in absolute measurements, but with a slightly longer span. If the wearer of our Kragelund garment were as tall as our 5'10" reference man, then a full-length version of the garment would have a 1.17 span:height ratio, right in the "slightly long sleeve" category.

So going back to our sleeve-style groupings, we can modify this to:

"normal" sleeve length (span:height ratio ca. 090-1.10)
- full-length garment: Bernulf & Italian alb #2
- short garment: Moselund, Skjoldehamn

slightly long sleeves (ratio ca. 1.14-1.17)
- full-length garment: Thomas alb
- short garment: Boksten, Kragelund

medium-long sleeves (ratio ca. 1.30)
- full-length garment: Italian albs #1, #4

extra-long sleeves (ratio > 1.40)
- extra-long garment: Italian albs #3, #5

So after we've normalized the span ratios to a hypothetical full-length garment, the "extra-long sleeves" category, instead of containing all short garments, contains all extra-long garments.

Another angle on evaluating whether a garment is full-length or shorter is to look at the ratio between the upper back (shoulder to waist) and the garment height. Since the majority of gussets begin around the natural waistline, a convenient first approximation for the upper back measurement is the distance from the shoulder to the start of the gussets. (Some of the Herjolfsnes garments clearly have gussets starting above the natural waistline, but I've avoided those in doing this analysis.)

Most of the Italian albs don't have the upper back length measurement, either because they shape is more of a tapered cone (with no clear waistline) or because the source has some measurements but not a diagram from which I can take the upper back dimension. The general proportions also suggest that the gussets may start lower when the garment itself is fuller through the torso in general (i.e., when the gussets aren't specifically needed to provide movement room). This looks like it may be the case for Italian alb #5 and possibly for #3 as well, but we'll come back to that.

The two full-length garments with upper back measurements (the Bernulf and Thomas albs) have a back:height ratio of 0.36-0.38. If we take the four shorter garments and normalize the back:height ratio to a hypothetical full-length garment, they fall in two groupings: the Moselund and Skjoldehamn garments with ratios of 0.28-0.30 (i.e., with the gussets starting proportionately higher on the body than the albs) and the Kragelund and Boksten garments with a ratio of 0.37-0.40 (i.e., fairly similar to the albs). Without normalizing, these garments fall in the same two groupings but the "alb-like" style has a ratio of 0.48-0.52 and the "higher gusset" style has a ratio of 0.35-0.36 (i.e., superficially similar to the alb type before normalizing). For the over-long albs, the initial impression of a "dropped waist" is borne out by the numbers. If we look at the actual proportions of the garments, the back:height ratios are 0.43 (for the shorter of the two) and 0.62 (for the really implausibly long one). If we normalize the ratios for our hypothetical reference man, the back:height is 0.55 and 0.88. These numbers tend to support the idea that much of the excess length was meant to be bloused up at the waist rather than dragging on the ground.

So lets go back and once again refine our type-groups:

"normal" sleeve length (span:height ratio ca. 090-1.10)
- full-length garment, back:height ratio ca. 0.36: Bernulf & Italian alb #2
- short garment, back-height ratio ca. 0.36 (but normalized to ca. 0.29): Moselund, Skjoldehamn

slightly long sleeves (ratio ca. 1.14-1.17)
- full-length garment, back:height ratio ca. 0.36: Thomas alb
- short garment, back:height ratio ca. 0.50 (but normalized to ca. 0.38): Boksten, Kragelund

medium-long sleeves (ratio ca. 1.30)
- full-length garment (no back:height ratio data): Italian albs #1, #4

extra-long sleeves (ratio > 1.40)
- extra-long garment, back:height ratio >0.40 (but normalized to > 0.50): Italian albs #3, #5

The least useful measurement for wearer height is the width of the garment at the shoulders (i.e., between the two sleeve seams) because it is driven both by overall size and by garment style. For the same arm span, this will be wider if the garment has unshaped armscyes (i.e., a dropped shoulder seam), and for garments with over-long sleeves, the length may occur either in the sleeves themselves or in fullness across the shoulders or both. The albs are normally cut with no shaping of the armscye (i.e., the shoulder measurement is simply the width of the main body panel) and are often relatively full through the body. The shoulder measurement tends to scale with the span:height ratio, that is, the albs with a "normal" span have a smaller shoulder measurement (24-33"), for those with a proportionately longer span the difference seems to come primarily from a wider shoulder measurement, in most cases 40-46". That is, we don't actually seem to be dealing with over-long sleeves but with more or less surplus fabric in the body. The odd man out is Italian alb #4 with a shoulder measurement of 28" but a span:height ratio in the medium category at 1.27. But this garment (when you look at the actual cutting layout) has side gussets that extend well up into the armscye, providing a great deal of the surplus fabric through the body. So functionally the dimensions correspond more to a garment with shaped armscyes.

The secular garments have narrower shoulder measurements, whether the armscyes are shaped or not and we find that not only are the panels with shaped armscyes slightly narrower (as expected) but there's a direct relationship between the normalized span:height ratio and the shoulder width (i.e., the "long" spans are due pretty much directly to the shoulder width). But in any event, there's significantly less fabric involved than for the albs: 19-22" for three of them, with the Skjoldehamn garment being the odd man out at 27".

The same data can be viewed from a different angle by considering the length of the sleeve (that is, the length of the piece of fabric used for the sleeve which, as noted above, may begin at or slightly lower than the point of the shoulder in anatomical terms). While the sleeve lengths range from 15" to 28"(!!!) the majority are in the 18-25" range. Within this range, items on the longer end of the range tend to be those garments that are generally "extra loose" (i.e., Italian albs #3 & #5) and the secular garments with at least slightly shaped armscyes (i.e., where the sleeves start higher on the shoulder) such as Moselund and Boksten. Italian alb #4 which, as noted above, effectively has shaped armscyes also has the longest sleeve length (28") so in this case is may be a combination of the shaped armscyes and simply being a generously cut garment. The Kragelund has slightly longer-than-expected sleeves, given that the body piece is unshaped at the armscyes, and this simply seems to be a quirk of the garment. (I made an exact-size reproduction of it and always have to roll up the sleeves at the wrist to make them manageable.) At the short end, the St. Thomas alb is the outlier at 15" -- although now that I look back at my sketchbook notes for this garment, the sleeve length isn't one of my original measurements and I'm guestimating from proportions, so I'd discount this one. As a general rule of thumb, the sleeve length tends to be closely similar to the width of the body panel or approximately half that of the body panel. I tend to suspect that this is not entirely unrelated to the tendency to cut body panels along the length of the fabric and sleeves across the width. That is, the relationships of the pattern pieces in this construction method may be due to evolving out of what works for particular standard widths of fabric rather than converging for some more theoretical reason.

At this point, we can go back to the Louis tunic. When comparing it to the above range of proportions, it matters significantly whether we use my estimate of the measurements or Burnham's. Burnham estimates the sleeves significantly longer than I do (22" to my 15") and the central panel of the tunic as slightly wider (22" vs. my 16") for a cumulative difference of Burnham's 66" span vs. my 46" span. Our height estimates are very close: Burnham's is approximately 43" and mine is 46". Our upper back height estimates are also different with hers 12" and mine 16" but the wording of her description sounds like she was simply assuming that the gussets would be taken from the off-cuts of the sleeves (making them the same length) while I was estimating them directly. So since the upper back length gets calculated from the total length minus the gussets, I'm not sure I'd put too much meaning in this difference if she didn't make an independent estimate. At any rate, the two sets of estimates come out as:

Item

Span

Height

Upper
Back

Shoulder

Wearer
Height

Span:Height

Back:Height

Length
Fraction

L(hrj)

46

46

16

16

?

1.00

0.35

?

L(B)

66

43

22

22

?

1.53

0.51

?




It may not be entirely coincidental that both sets of estimates have identical upper back height and shoulder width estimates -- relative proportions are often more obvious (and therefore more likely to be corrected for) than absolute dimensions.

Looking first at my dimensions, if we allow the assumption that sleeves will never be shorter than wrist-length (and in this case are highly unlikely to be over-long), then since the actual span:height ratio is 1.00, the garment must be very close to full-length for the wearer. The back:height ratio fits with that for the full-length albs (but also that for some of the shorter secular garments). The slightly shaped shoulder width is narrower than what is typical for albs but perhaps proportional for the overall smaller size. The sleeve length is closely similar to the shoulder width although both are at the low end of the range for the garments surveyed. Now, if the 46" height is "full length", then the wearer would be expected to be 4'6" tall. For an adult man, this is shorter than the definition of a midget. Since nobody seems to have described St. Louis in those terms, then a garment with these dimensions either would not have been worn by him or would have been worn by him when he was a child.

Before we conclude that this means I must be horribly wrong in my estimate of the garment's dimensions and construction (never mind that other people have looked at the original and come to similar conclusions), let's look at Burnham's version.

Burnham's dimensions are exceedingly similar in all respects to those of the Kragelund and Boksten tunics.

Item

Span

Height

Upper
Back

Shoulder

Wearer
Height

Span:Height

Back:Height

Length
Fraction

L(B)

66

43

22

22

?

1.53

0.51

?

K

70

46

24

22

*

1.52

0.52

*

Bk

67

46

22

19

69

1.46

0.48

0.78



This would make it a reasonable calf-length garment on our reference man of 5'10" height.

Well that's quite a difference. In defense of my estimates, having started by assuming that Burnham's pattern was accurate (and having made and worn a copy based on her proportions) I was struck forcibly enough by the shorter proportions of the sleeves when I saw the original that I made special note of it. Furthermore, the thread-trace analysis clearly indicates that Burnham is wrong (or at least vastly over-simplifying) in portraying the body panel as a simple rectangle with minimal shaping at the armscye rather than being tapered generally and strongly shaped at the armscye.

But when it comes down to it, the most important difference in the interpretations when determining the size of the wearer is the span dimension. And this difference derives primarily from differences in the length of the sleeve pieces themselves (14" total) with a smaller amount from the width of the shoulders (ca. 4-6") and Burnham doesn't actually provide a dimension for the sleeve length -- the length is taken from the relative proportions of the layout diagram. (Only the fabric width is given explicitly.) And the sleeve length and relative proportions are the parts of my estimates that I'm most confident about.

So I'm sticking with my dimensions and the conclusion that the St. Louis tunic was not worn by a normal-sized adult man. Which is not necessarily to say that the association with St. Louis is wrong, but in the absence of evidence that he was strikingly short in stature, it could only have been worn by him when he was a child.

Comments

( 2 comments — Leave a comment )
hudebnik
Sep. 6th, 2010 12:38 pm (UTC)
It seems to me another useful measurement would be across the chest just under the arms (and gussets, if there are any). This could be fairly large relative to body size if the garment was loose, but it cannot be too small relative to body size, so a ratio with it puts an upper bound on height.
hrj
Sep. 6th, 2010 05:36 pm (UTC)
I considered that one, but there are two problems with it. For one, because so many of the garments of this general type are relatively loose across the torso/chest/belly, it would be hard to identify a benchmark for just how much fullness is the minimum necessity (since it's always going to be more than the anatomical measurement). But also, it would be very hard to identify a standardized method of taking the measurement. As a general rule, the narrowest point is going to be below the attachment of the sleeve+sleeve-gusset and above the start of any skirt gussets ... but depending on the construction this point can either fall across a broad band from armpit to point of the hips, or it may not exist at all (if the sleeve attachment is very deep, as it is for some albs).

I was worried enough about the shoulder measurement being more noise than signal, except that it helped provide a reality-check on where the overlong spans were coming from. I think I may go back and add in a metric for sleeve-length vs. height, since the shoulder data suggests that the body looseness is the primary driving factor for long spans. If I were dealing with more garments with strongly shaped armscyes then I'd have to worry about the difference between dropped and normal shoulders, but for the set of garments I stuck to here, there's relatively little shaping.

One thing I didn't mention in the above analysis is that at one time I starting making up an exact-size reproduction of the garment based on my article ... and rapidly noticed the same sizing issue in a very practical way. Hmm, it occurs to me that another non-invasive evaluation of the interpretation would be to take an exact-size reproduction to Paris and hold it up in front of the display case ....
( 2 comments — Leave a comment )