An Analysis of Allen Collection M-2

Mart Shearer

Over June 21-22, 2014 at an Armor Study Session hosted by Wade Allen, he, Tom Biliter and I had opportunity to examine a mail shirt, collection number M-2 in some detail. Thanks are due to both of these men for their invaluable help.


M-2 is a mail shirt of all-riveted construction which has half sleeves without tapering, no collar, and knee length skirts with central splits in front and back. Although it had been assigned a likely Indo-Persian origin and recent dating, full analysis bring that assignment into question. I believe the construction and tailoring also give clues as to the provenance and allows a tentative dating, which is troubling.


The body has some repair rings, but is generally in a good state with smooth rings and a bright appearance. The main body of the shirt is composed of fairly uniform riveted rings of round section drawn wire. These rings have a flattened overlap, right over left, which appears to have been cut with shears. The bottom of the overlap has a gentle curve, while the rivet bulge has a graceful arc toward the small rivet head. No sign of a watershed is seen at the rivet. Erik Schmid has reported the watershed effect to be a likely sign of German manufacture, but none are present throughout M-2. 6 rings were measured with a digital micrometer from the front, back and side. These appear relatively uniform, with wire thicknesses having an observed range of 0.95 mm to 1.14 mm and an average of 1.07 mm. Ring outer diameter measured parallel to the rivet join have an observed range of 7.2 to 7.8 mm, with an average outer diameter of 7.53 mm. No observable thickening is seen in the rings of the chest compared to the back or side. Several missing rivets were observed, where the shape of the drift used to make the hole could be determined on the upper edge of the ring. This appears to be an elongated pentagon with long base and short sides.

Importantly, the body is seen to be composed of a number of patches which are connected with larger, flatter riveted rings.

Neck Opening

The neck opening is square, with a small vertical slit at the front of the neck. The base of this slit is woven into a slight overlap, right over left, about five rings wide. This is clearly designed to help minimize any gapping at the opening after presumably lacing it shut. The neck opening is lined with a fragmentary row of smaller riveted mail rings, 3 rows wide, applied like bias tape on a fabric edge. These rows continue in the same direction regardless of the direction of the mail in the shirt body, requiring a 90 degree join along the sides of the neck hole. Two rings were measured, and the outer diameters were determined to be 6.8 mm and 6.5 mm with thicknesses of the round section rings being 1.05 mm and 1.01 mm.


The knee length skirts are split front and back their entire length. The skirts are attached to the body with the same sort of flattened rings seen connecting the body patches. Although the rings in the skirt are noticeably of lighter construction than the body, their basic form and shape of riveted join are identical to the rings in the body. Four rings, two from each side, were measured. Rings were observed to have an outer diameter range of 7.56 mm to 8 mm, with the average observed size being 7.82 mm. The round wire rings have measured diameters ranging from 0.85 mm to 1.08 mm, with the average being .96 mm, making the skirt rings slightly thinner and larger than the more critical defense of those in the body.


The half sleeves are only about elbow length or slightly longer. There are no signs of tapering in the sleeves to narrow them. The sleeve is seamed up on the inside line with the same flattened rings used to connect body patches and skirts. The sleeves are composed of noticeably lighter wire which has been slightly flattened and formed into rings with a slight oval shape. All riveted construction is used throughout them, but the lighter construction has led to more losses, and the sleeves have more holes than the body and skirt. The sleeves appear to be joined directly to the body without the flat rings used in other connections, though a few of those flat rings appear, possibly as repair links, or because the line of the connection rings has been disrupted by the loss of rings.

Four rings were measured, two from each sleeve. wire thickness seems more variable, with a range of thickness being 0.56 mm to 0.77 mm, and wire width after flattening being 0.95 mm to 1.06 mm. The average cross section for the flattened wire in the sleeves is 0.67 mm x 1 mm. Round wire giving a similar are would be approximately 0.82 mm in diameter. The oval shaped rings were measured parallel to the rivet joint and found to have outer dimensions ranging from 7.87 mm to 8.12 mm. A second measurement taken across the riveted joint to the other side had ranges of 7.28 mm to 7.81 mm. The average size were outer diameter #1 being 7.97 mm and outer diameter #2 being 7.47 mm.

Connecting Rings

As has been noted, the shirt is assembled from sections using larger, flatter rings as connectors. These rings generally appear in straight horizontal and vertical rows and lines. A few isolated ones appear in the main fabric of the shirt and might be repairing flaws in the normal fabric. Four of these rings were samples from different seams. The rings are of flattened section with thickness ranging from 0.76 mm to 0.96 mm and width measuring between 1.16 mm and 1.48 mm. The average shape of this flattened wire is an ellipse measuring 0.85 x 1.32 mm. Like the flattened sleeve rings, the larger connecting rings have a slight oval form Outer diameter #1, measured parallel to the join has a range of 9.46 mm to 10.06 mm, while outer diameter #2 measured over the rivet has a range of 8.74 mm to 9.41 mm. An average connecting ring has an oval form measuring 9.75 mm by 9.07 mm. These larger rings are easy to spot in the in the body of the shirt with it's rings in the 7.5 to 7.8 mm outer diameter range.


The tailoring of the shirt seems to be accomplished with thinner rings resembling those used in the sleeves. There appears to be no shaping to the sleeves: They are simple tubes seamed together with the connecting rings at the inside edge. The body of the shirt appears to have had the common expansions beginning around the shoulder or collar bone and enlarging towards the shoulder blade. These aid in the forward motion of the arm. There are also two vertical lines of expansion in the lower back and over the buttocks. This may have accommodated some fashion or the personal physique of a man with a big bottom. (Some jokes were shared over this.) The body of the shirt is pieced together on the left side from multiple patches. It was suggested that this was a later modification, perhaps to enlarge the shirt. This idea was rejected since the rings inside the seamed patches is well matched to the rings in the body on the other side of the seam. The lowest of these patches exhibits some bagginess.

It seems likely that different men in the same shop were responsible for differing aspects in the construction. Perhaps one made the main straight patch which composes the body, while another made sleeves and added expansions and contractions for the first man. Finally the master pieced these sub-assemblies using another style of ring. If the same size and style of ring had been used to form the connections, the seams would be indistinguishable from the rest of the fabric. If mail had been bought like fabric off of a bolt and then cut to fit, we would have found mor connector rings around the tailoring expansions.

Dating the Haubergeon

Early European and almost all "quot;Indo-Persian"quot; mail seems to have been made of demi-riveted construction, i.e. half solid and half riveted rings. European inventory records from the late 13th and early 14th centuries begin distinguishing between this form of construction and fully riveted mail. The fact that this haubergeon is of all-riveted construction points to a European origin. While all riveted construction may appear in some 15th or 16th century Turkish mail shirts, none of them use wedge riveting. While not all European shirts are wedge riveted, the all riveted construction combined with the wedge riveting almost certainly points to a European provenance.

The style of riveting could easily be a late 14th or early 15th century form, but the tailoring of the shirt with its long split skirt points to an earlier dating, in my view. Knee length hauberks and haubergeons with long split skirts are often referred to as "quot;Norman hauberks"quot; due to their similarity to the Bayeux Tapestry depiction of such armors being carried on poles, but early European mail seems to have been demi-riveted. Knee length mail can be found on effigies and in manuscript miniatures into the 1340s; however, many of these examples keep us from observing the length of the skirt slit due to the long surcoats worn above mail. Most examples have fitted long sleeves with attached mittens. Wide elbow length sleeves to start to become noticeable in the decades from 1320-1340.

I am incredulous that this is an haubergeon dating to the first half of the 14th century in Europe, but this seems to be the way that the evidence points. Perhaps there are later examples of such long skirts in a European context which would allow a later dating.