This page last modified: 27 December 2014 (commentary about textual transmission added). This section is very much "under construction".
A note on copyright and the Notitia
These web-pages illustrate these shield patterns (for how we can be sure they are shield patterns, given they are not labelled as such, see my discussion here). Several manuscripts and other documents containing copies of these pictures exist; as such, they show various differences. The pattern shown on this website are taken from:
1. the manuscript (manuscript "O") that now resides in the Bodleian Library in Oxford. The versions on this site were originally derived from scans taken by Nik Gaukroger, but are now supplemented by the Bodleian's own scans, many of which are now available on-line);Other surviving manuscripts and fragments either do not contain any sections showing shields, such as the ten surviving pages (1r, 1v, 2r, 2v, 3r, 3v, 4r, 4v, 5r, 5v) of the Cambridge manuscript, manuscript "C", now residing in the Fitzwilliam Museum, Cambridge (but see here); or they omit the shield patterns, albeit sometimes with spaces and labels in place for them to be added, such as the Vatican manuscript, manuscript "V"; or they are copies of one of the sets of pictures above, and thus have little independent value, such as the Madrid manuscript m36 (copied from O); or simply have not been scanned and made available on-line, and thus their pictures are currently unavailable, such as the Trento manuscript W3101, manuscript "T", or that of the Victorian and Albert Museum, MSL/1957/2668, manuscript "L".
2. the manuscript (manuscript "P") that now lies in the Bibliotheque nationale de France, Paris, and which is now also available on-line, at the library's web-site (sorry about the lack of a grave accent on the e in Bibliotheque!);
3. the manuscript that now lies in the Bavarian State Library, Munich, and which contains two sets of pictures (manuscript sets "M" and "W"), and which are also available on-line, at the library's web-site;
4. the Froben printed "edition" of 1552 (imprint "B"); at least two versions of which are available on-line: that of the University of Cologne unfortunately does not have colour pictures, whereas that of Gent University, and digitised by Google books, does; and
5. the mostly-lost Frankfurt manuscript (manuscript "F"), of which a single leaf ("Ff") now resides in the Goethe University, Frankfurt am main, and which is available on-line; this one leaf just so happens to show some 60 shield patterns of western auxilia palatina. Two more leaves (manuscript "Fl") now reside in the Leiden University library, but they are of sections that have no shield patterns (the Dux Mesopotamiae section is available on-line).
The pictures, despite apparently all having been copied, directly or indirectly, from the same now-lost manuscript, vary considerably in style and execution. I give a more extended discussion here, but in brief:
P: The P manuscript's pictures are probably the clearest (at least, as currently available on-line), but have a relatively high incidence of spelling mistakes in the labels, and occasionally omit a detail that may be found elsewhere.In general, no one set can be considered "the best"; each must be evaluated on its own merits with regard to any particular shield pattern, which is why I am presenting them together.
W: Those of W are rather crude, especially with regard to proportions; they also have problems with fading colours. However, W often preserves details, albeit crudely, that may be skipped in one or more of the other manuscripts (see more below).
M: Those of M have even more problems with colours, sometimes miss details, and in addition those details that are present are often over-embellished. These were the patterns Otto Seeck presented in his widely-consulted 1876 edition.
B: The patterns of B, a printed imprint of an apparently no-longer existing manuscript copy once residing in Basel, are, if anything, even more fanciful than M. Further, its illustrations are very frequently reversed, and are often displayed in the wrong order; nonetheless, they do occasionally show a significant detail lost elsewhere, such as with the shield ascribed to the eastern Mattiaci iuniores.
Ff: Those surviving from Ff have not only very faded colours, but are are rather damaged, some very much so, and are also over-embellished, but their proportions often look "right".
O: Those of O are generally a happy medium, and are executed to a higher degree of quality than those of P; unfortunately the available pictures of the illustrations are of limited resolution.
Note that although the surviving manuscripts all derive, directly or indirectly, from the same now-lost manuscript, that was once housed in a library attached to Speyer Cathedral, that manuscript, the Codex Spirensis, was not the Roman original. The Codex Spirensis was apparently itself a copy, and not necessarily a direct copy, of a Carolingian manuscript which was a copy, and again, not necessarily a direct copy, of "the" original Roman exemplar (which may have been a compilation of more than one document). So when I talk about "the" "original" Roman Notitia, you should understand that I am speculating about its contents based on inferences drawn through several layers of hand-copying, with the inevitable attendant mistakes.
Much of the Notitia text that accompanies the pictures, such as the names of the units, was originally taken from Seeck's 1876 edition, as presented by Halstein Sjolie at his web-site; I have noted some changes in certain places, particularly where they impact the identification of units with certain shield patterns: Seeck often amended a unit label for no good reason, and sometimes amended it is such a way to actively hinder understanding (e.g. with the Legio secundae britannicae siue secundani).
Only some of the units mentioned in the Notitia have their shield patterns depicted. In particular, none of those of the limitanei or "border" troops are illustrated, nor (with one exception) are those of the eastern cavalry units, hence the complete lack of shield patterns for some of the forces listed below. Why this is so is an unanswered question. It may be that a Carolingian copyist grew tired of drawing so many patterns, and yet the document probably owed its continued existance in part because of the visual interest these very same patterns provided. But the selectivity of the omissions is striking: for example, amongst the infantry of the Magister Militum per Illyricum, it is the pseudocomitatenses units (i.e. recently-transfered limitanei units) - and only the pseudocomitatenses units - that are not illustrated. It appears quite likely that the "original" Roman compilation simply did not contain the "missing" shield patterns. For example, while the other eastern Magistri's pages show up to two dozen patterns each, that of the Magister Militum per Illyricum shows only 15 in the same space, so there was certainly room enough to illustrate the "missing" 9 pseudocomitatenses units.
The Eastern empire:In addition to these major blocks of illustrations, many units of the units' patterns are shown and discussed individually: you can reach them from links from the above officers' lists, or you can look for them alphabetically.Magister Militum Praesentalis I - 120 versions of 24 shieldsThe Western empire:
Magister Militum Praesentalis II - 119 versions of 24 shields
Magister Militum per Orientem - 105 versions of 21 shields
Magister Militum per Thracias - 105 versions of 21 shields
Magister Militum per Illyricum - 75 versions of 15 shields
Magister Officiorum - 30 versions of 6 shields
Comes domesticorum equitum, peditum - 10 versions of 2 shields
Comes limitis Aegypti - 3 shields taken from elsewhere in the Notitia
Comes per Isauriam
Dux Libyarium (only the command's heading remaining, and in only some manuscript versions)
Dux Thebaidos - 4 shields taken from elsewhere in the Notitia
Dux Syriae et Eufratensis Syriae
Dux Armeniae - 5 versions of 1 shield taken from elsewhere in the Notitia
Dux Moesiae secundae
Dux Moesiae primae
Dux Daciae ripensis - 10 versions of 2 shields taken from elsewhere in the NotitiaMagister Peditum - 674 versions of 123 shields
Magister Peditum's Italian command - 43 shields taken from elsewhere in the Notitia
Magister Equitum - 194 versions of 39 shields
Magister Equitum's Gallic command - 46 shields taken from elsewhere in the Notitia
Magister Officiorum - 35 (or 38) versions of 7 shields
Comes domesticorum equitum, peditum - 10 versions of 2 shields
Comes Illyricum - 21 shields taken from elsewhere in the Notitia
"Comes" Hispenias - 21 shields taken from elsewhere in the Notitia
Comes Africae - 31 shields taken from elsewhere in the Notitia
Comes Tingitaniae - 32 versions of 6 shields taken from elsewhere in the Notitia
Comes litoris Saxonici per Britanniam - 5 versions of 1 shield taken from elsewhere in the Notitia
Comes Britanniae - 4 shields taken from elsewhere in the Notitia
Comes Italiae - 3 shields taken from elsewhere in the Notitia
Comes Argentoratensis - 5 versions of 1 shield taken from elsewhere in the Notitia
Dux et praeses provinciae Mauritaniae et Caesariensis - 10 versions of 2 shields taken from elsewhere in the Notitia
Dux Provinciae Tripolitanae - 5 versions of 1 shield taken from elsewhere in the Notitia
Dux Provinciae Pannoniae secundae ripariensis et Saviae - 37 versions of 7 shields taken from elsewhere in the Notitia
Dux Provinciae Valeriae ripensis
Dux Pannoniae primae et Norici ripensis - 45 versions of 9 shields taken from elsewhere in the Notitia
Dux Raetiae primae et secundae
Dux Provinciae Sequanici - 11 versions of 2 shields taken from elsewhere in the Notitia
Dux tractus Armoricani et Nervicani - 6 shields taken from elsewhere in the Notitia
Dux Belgicae secundae
Dux Germaniae primae (section apparently deleted: no towns, officers or troops allocated)
Dux Britanniarum - 16 versions of 3 shields taken from elsewhere in the Notitia
Dux Mogontiacensis - 20 versions of 4 shields taken from elsewhere in the Notitia
As mentioned above, the manuscripts from which these illustrations are taken do not derive from the late Roman period: the only surviving copies date from the 15th and 16th centuries, and all derive from a single codex - the Codex Spirensis - that is known to have existed in 1542 but is now lost (and has been lost since the 17th century), and which was itself derived from a possibly 9th century Carolingian copy and not the Roman original. The reliability of even the "original" Roman document has been questioned, and in particular, its purpose, although A. M. Kaiser has recently (2014) conducted an analysis of the papyrological evidence for the Egyptian units mentioned in the Notitia: "Egyptian Units and the reliability of the Notitia dignitatum, pars Oriens", available here, which demonstrates the reliability of the Egyptian listings, and by inference, those of other (eastern) sections of the Notitia. Nonetheless, the reliability of the textual listings and the accompanying pictures are two different things, and this raises the question of the authenticity of the illustrations: are they faithful reproductions of original Roman material; wholesale fabrications of a medieval scribe; or something in between?
Robert Grigg was the first to raise serious questions about the authenticity of the Notitia patterns, in a couple of articles in the Journal of Roman Studies (Vol. 69 (1979), 107-124, and Vol. 73 (1983), 132-142, respectively).
We can at least be sure that at least some of the patterns shown in the surviving copies of the Notitia are not completely wholesale medieval fabrications because the patterns shown are in general terms very similar in style to other late Roman illustrations, and in some cases, essentially identical (see my discussion here). But that doesn't mean they all are. Grigg made much of how the shield patterns shown appeared to get simpler and simpler the further into the manuscript one delves, as if the illustrator got tired of inventing new designs. But even if this is the case (which I would partially dispute - what Grigg saw as "essentially blank" often looks nothing like that to me) - there may be one or more perfectly reasonable explanations for this. For example, it appears that some of most senior units (i.e. those depicted first, in order of precedence) tend to have the most complex shield patterns, at least in the east, as if accumulating "badges" over time. Since the patterns for the least senior units in the East (i.e. the 2nd quarter of the document) are missing, this alone could account for a diminishment of pattern complexity.
Grigg also questioned the symbolism involved. For example, only a couple of shield patterns appear to illustrate a wreath, even though this was a very common shield motif in former times; further, in each, the depiction is crude, like a circle; this made Grigg suspect other wreathes may have been illustrated, but had been stylized to the point of appearing as circular bands (I agree this has likely occurred, although possibly as much for reasons of scale as lack of care in copying). And thunderbolts, so ubiquitous in the past, are now seemingly entirely lacking. But there are also arguments as to why this might be so, even without copying problems (which are, of course, inevitable in transmitting documents by hand). Symbols move with the times after all. Thunderbolts hurled by Jupiter may simply have been politically incorrect imagery in the newly-Christianized post-Constantinian world in which the Notitia was first drawn up.
Grigg was on stronger ground when he questioned the faithfulness of individual patterns, in other words, whether the identification of any one particular pattern with any one particular unit is secure. This is because there is strong evidence that at least some of the patterns may be misaligned with their corresponding units. For instance, on the first page of the Magister Peditum's patterns, the Mattiaci (iuniores) and the Ascarii seniores, neighbouring entries, are given essentially the same pattern. It looks like one has been accidentally duplicated. Maybe one of the two is just missing, leaving the rest unchanged; or maybe all subsequent entries have to be shifted over one place (this was noticed at least as long ago as by Seeck). But there are further problems too. For example, only one Equites Brachiati is illustrated and listed as part of the Magister Equitum's cavalry roster, whereas both an Equites Brachiati seniores and an Equites Brachiati iuniores are listed as being assigned to field commands, and so there is no way of conclusively knowing which of these two units the shield pattern labelled Equites Brachiati belongs to.
Having said this, a close examination of the patterns shows in many cases clear "family" resemblances. In some cases, units that are known to have historically been connected with each other by citations from literary texts, such as the historian Ammianus, bear strikingly similar patterns. And in other cases, units with similar names bear very similar patterns, for example. Grigg downplayed this in part, because there were many instances in which "expected" relationships were not in evidence. But that is begging the question. There is no need to assume every "related" set of units must perforce have had similar shield patterns just because some of them did. And to say just because a unit named Heculiani shows an eagle pattern, instead of depicting the god Hercules, that it must perforce be inauthentic, is very questionable logic indeed.
Perhaps the following extract best sums up Grigg's attempts to deny obvious relationships between the shield patterns shown and the unit titles they are attached to. Regarding the following four pairs of units: the Ioviani iuniores and the Herculiani iuniores; the Fortenses and the Nervii; the Daci and the Scythae; and the Primani and the Undecimani, Grigg stated (his 1983 paper, page 138) that "in only one of the four pairs of units (Not. Or. vi, 5, 6, the Primani and Undecimani) are the titles even remotely related". I simply can't see how someone can deny the names Jupiter and Hercules are related, given they represent a father-son pair. Likewise the names Daci and Scythae are related through being the names of neighbouring provinces, and the names Fortenses and Nervii are related through both having the same meaning, of "steadfast".
Grigg also made much (his 1983 paper again, page 140) of how units that bear similar titles are much more likely to have similar shield patterns only when in the same command, which he said can "best be explained by presuming that the artist's concern for the co-ordination of the emblems of related units was more easily defeated when time and energy were required to satisfy it". Clearly the thought that units might have received new shield patterns when they were transferred to a new command never occurred to him. Most of Grigg's arguments are unfortunately similarly lacking in rigour (or perhaps, fortunately lacking, given if they were true, the document would less useful as an historical source).
So we need to distinguish between various levels of "authenticity": is a particular pattern in the Notitia merely "an authentic shield pattern of some late Roman unit", or "the authentic shield pattern of the particular late Roman unit that it is labelled as belonging to..."
Furthermore, even assuming they are "authentic", the patterns shown can't be claimed to be entirely faithful reproductions (as in "true to the real-life prototype"), even individually, as there are in some places considerable differences between the various surviving manuscript illustrations - differences that these web pages illustrate. Obviously there are corruptions that must inevitably be introduced by the multiplicity of copying events between the 5th century original and the surviving manuscripts which leads to discrepancies between the surviving versions.
Indeed, the reason the Munich manuscript contains two sets of pictures is that the manuscripts' owner wanted not only a "new" set of manuscript pictures, but deliberately sought out an "old" set as well, that best preserved the form of the Codex Spirensis illustrations. The first ("new") set of pictures, manuscript copy "M" according to Ingo Maier's scheme) look prettier than the second set (copy "W"), and while the second are rougher in execution, they appear to be more faithful in many ways to the Codex Spirensis and thus to the lost Roman exemplar. For example, in W, one of the helmets shown under the western Magister Officiorum looks - admittedly vaguely - like a fairly normal late-Roman conical-type helmet of the type that would later be called a "spangenhelm", complete with crest and what look to be singly-scalloped cheek guards. But in M, this crude picture has been transformed into a much clearer but easily recognizably anachronistic German sallet, albeit embellished with a crest and two tassels. Prettier, but clearly less authentic. Likewise the "castles" in W are rough hexagonal forts, whereas in M they have been transformed into late medieval European castles and walled towns.
The Notitia Dignitatum as it has come down to us is apparently divided into halves pertaining to the western and the eastern portions of the Roman empire (or rather, eastern and western portions; the Romans not adhering to the same worldview as us). On internal evidence, the eastern section appears to be (at least at first sight) an essentially uniform composition dating to (or heavily revised in) ca. 394 AD; some would date almost all of it to ca. 400 AD, however. The western section seems to initially date from the same time, and then having evidently undergone considerable revision, with some items seemingly dating to 420 AD or even as late as 425 AD. (I say "apparently divided" into two halves, because part of the perceived division is due to Seeck's rearrangement of the material, and is not reflected in the same way in the actual manuscripts themselves).
Thus "the Notitia" does not present a snap-shot picture of the Roman army, but rather a view of many changes occuring over 30 or so years, and disentangling these is a tricky business at best. Many of the units listed in the western section for instance seem to be duplicated in two (or even more) different commands. It is difficult to say which is the 'original' posting and which is the 'latest' one (assuming, that is, we are not dealing with two identically named units; perhaps because they are two halves of a unit split between different stations that have not yet acquired a unique identity!). Nor is it easy to discern if a given unit in a field army with an identical name to a unit in a limitanei area is merely a detachment of the latter, or whether the unit in its entirety was withdrawn from the frontier...
In addition to displaying the shield patterns, I have endeavoured to give some sort of commentary about the units and army concerned. I make no pretence of being an expert in the area however; take my conclusions with a grain of salt. Some comments derive from (or argue against) AHM Jones' analysis of the Notitia Dignitatum in "The Later Roman Empire, 284-602. A Social, Economic and Administrative Survey" by A.H.M. Jones, The John Hopkins University Press, 1986, ISBN 0-8018-3285-3 (this is the single-volume paperback edition; sometimes I quote the separate hardback volumes). References to 'Burns' are to "Barbarians within the gates of Rome" by T.S. Burns, Indiana University Press, 1994, ISBN 0-253-31288-4. Various other works are referenced when mentioned. Some of the epigraphical evidence for British units comes from the valuable Roman Inscriptions of Britain site and much other British information from the excellent www.Roman-Britain.org site. A potted history of the old legions in previous centuries can be found at www.livius.org (be aware that many things stated here as apparent facts are mere suppositions, however). Evidence is somewhat biased towards British units not least because I am an English speaker and can't access other languages so easily, but also because British archaeological comparisons are comparatively extensive.
I have compiled an alphabetical list of the units mentioned in the Notitia which may help people in analysing the distribution of units, etc. Some include links to their shield patterns, etc.
Some troop type statistics and analyses are described below:
Singulares, Domestici, and Bucellarii
Return to my Ancient Military History index page.