Dillying and Dallying with Watermarks
A distinguishing mark or device impressed in the substance of a sheet of paper during manufacture, usually barely noticeable except when the sheet is held against strong light.
The Oxford English Dictionary, s.v. Watermark § 5
Watermarks like wrens go in pairs.
Allan H. Stevenson, ‘Watermarks are Twins’, Studies in Bibliography, vol. 4 (1951)
A daring dash of schoolboys, safely, shoulder to shoulder, with their fathers’ trilbies cocked at a desperate angle over one eye, winked at and whistled after procession past the swings of two girls arm-in-arm: one pert and pretty, and always one with glasses.
Dylan Thomas, Quite Early One Morning (1954)
Not surprisingly, the Guide’s graphically enticing description of the general state of affairs [the Fuolornis Fire Dragons] on this planet has proved to be astonishingly popular amongst hitch-hikers who allow themselves to be guided by it, and so it has simply never been taken out, and it is therefore left to latter-day travelers to find out for themselves that today’s modern Brequinda in the City State of Avalars is now little more than concrete, strip joints and Dragon Burger Bars.
Douglas Adams, So Long and Thanks for All the Fish (1984)
What can be said about watermarks?
That they are beautiful, enigmatic, intriguing, fascinating … (did I just describe myself? well, like Garfield, whom I fervently admire, if I were just a little bit more modest, I’d be perfect).
People who know nothing about paper are often filled with wonder when they see these translucent images of long-lost or fantastic objects peering out of centuries-old sheets of paper.
Of course just being able to recognise a watermark is an underappreciated skill that requires in the first place a more than passing acquaintance with the visual conventions of filigranology, since what appear to a novice a flying saucer and a space-rocket are in fact a cardinal’s hat and a key (at least, one hopes, but UFO-ologists may think differently).
Matters are complicated, especially in printed books, by formats, bindings and long-standing conservation practice (i.e. what goes under the iniquitous and misleading term of “book restoration”), to the point that even finding the sign can be a frustrating experience. Rare books in prestigious collections have often been washed and ironed before rebinding, with consequent loss of evidence; any format lesser than folio pushes the watermark into the margin or onto the edge of the leaf, where chunks can be cut away; heavy manuscript or print can obscure the forms of the object represented; and the ubiquitous neon light of expensively refurbished reading rooms in libraries leaves paper about as transparent as marble. And librarians can be extraordinarily unhelpful (though most are absolutely charming!).
In what are therefore far from ideal working conditions only genuine expertise can piece together the disiecta membra, give a name to the object and point the inquirer to the relevant page in Briquet or some other manual. But there is an touch of vicarious satisfaction, when others are not even sure that something is there, in being consulted in some library, glancing at the paper stock, and recognising a fragment of a foot as belonging to a Pilgrim or to an Angel. (Well, even if you are not insanely into filigranological one-upmanship, it’s usually worth a coffee and a brioche from a fawningly grateful colleague.)
Nothing like watermarks exposes the bitter fragility of observation. It is not gracious to say so (so I’ll say it), but few, if any, scholars, even experienced bibliographers, can be trusted to observe paper in general and watermarks in particular with any degree of reliability or accuracy. They fail to take notice of the size of the sheet or the distance between chain-lines, whether the watermark is resting on a supplementary chain-line, whether they are looking at the sheet from the mould or the felt side, whether the watermark is in the right or left-hand side of the mould, whether they have identified the twin watermarks, whether there are tranchefiles, and so on and so forth. These drawbacks in preparation and knowledge on the other hand rarely, if ever, prevent them from building towering castles of hypothesis on these sloppy, wobbly, happy-go-lucky foundations.
If a casual inquiry about whether they have understood any of the above points meets only with a blank look, the wisest policy is just to walk away. Some people will never get it.
Watermarks. The Earliest Dates
A sheet of paper is a moulded object. Briquet said it. Please understand this one, single, bloody-minded fact.
If we can work backwards from the object made on the mould to establish the physical characteristics of the mould itself, and thus to relate it to other objects formed on the same mould, the outcome is an important tool for bibliographical research.
On this particular fact I have always felt that the man himself, who after all spoke in the guise of a former employee of a paper-mill and somebody who worked all his life in the stationary trade, so he had a good idea of how often it was necessary to fork out for a new pair of moulds, puts things beautifully in a rightly celebrated passage, first published in 1892 and reprinted in his Opuscula in 1955 . Harken therefore:Toute feuille de papier filigrané porte en elle-même son acte de naissance, le difficile est de le déchiffrer. Rappelons qu’une telle feuille a reçu en effet l’empreinte de la forme sur la quelle elle a été faite; c’est donc un objet moulé, comme une médaille ou une monnaie, dont tous les exemplaires sont semblables entre eux. Or, une forme à papier est promptement mise hors de service; sa durée moyenne ne dépasse pas deux ans. Lorsqu’elle est usée, elle est remplacée par une autre, qui n’est jamais absolument identique à la précédente; elle en diffère par la vergeure, par le nombre et l’écartement des pontuseaux, par les contours ou les dimensions du filigrane ou par la position qu’occupe ce dernier sur la forme. Pour pouvoir préciser la date de fabrication d’une feuille de papier, il ne suffit donc pas qu’elle porte un filigrane analogue à celui d’un papier d’une date connue; il faut que les deux filigranes soient identiques, placés au même endroit de la forme, il faut que le format, la vergeure et les pontuseaux des papiers comparés soient les mêmes. Il convient encore de rappeler que, dans la fabrication du papier, on se sert toujours simultanément de deux formes et que, bien qu’exactement contemporaines, ces deux formes offrent toujours quelque dissemblance.
Translation: Every sheet of watermarked paper is in itself its own birth certificate. The difficulty is in deciphering it. Remember that every such sheet bears the imprint of the mould on which it was made. It is therefore a moulded object, like a medal or a coin, of which all the copies are alike. Now, a papermaking mould does not last long, on average not more than a couple of years. When it is worn out, it is replaced by another one, which is never absolutely identical to the previous one; it will differ in the wires, in the number and the distances between the chain lines, by the shape and the size of the watermark or by the placing of the same on the mould. In order to be able to state the date of fabrication of a sheet of paper, it is not enough therefore that it has a watermark similar to that on a dated piece of paper; the watermarks have to be identical, positioned at the same point on the mould, and the sheet-size, the wires and the chain lines must also be the same. It should be remembered moreover that, when making paper, two moulds are used at the same time and therefore, although made and shaped simultaneously, these two moulds always present some differences.
Of course this statement needs some qualification. Pairs of moulds made for larger sizes of paper were used much less and thus lasted much longer (sometimes decades rather than years); stocks of paper could be warehoused for long periods, and so on and so forth, but when all is said and done the above is a spot-on summary of how paper scholars think and go about their business.
Not all watermarks are equal, although some are more equal than others. Over a period of seven centuries watermarks go through several evolutions in terms of their placing and their functions, which vary according to the different geographical areas.
The earliest dated instance of a watermark, according to most people who have written about paper since Briquet’s magnum opus first appeared in 1907, is in a document of “1282” in the city archive at Bologna (Les filigranes, n. 5410). The year 1282 forms part of the title of the work, which was perhaps a shade audacious, given that it has had the unfortunate and perhaps unintentional consequence of canonizing the statement .
In fact, the evidence is rather more dubious and the correctness of the date is very much open to question. If we actually take the bother to read entry n. 5410, which refers to a tracing of a large Greek cross, i.e. in a circle with four pommels, it should be noted, first, that the date was followed by a question mark, so Briquet was not sure about it – what he writes is “Bologne, 1282? et 1287-88”, followed by references to another larger sheet found at Naples, dated 1321, and to Keinz’s 1896 catalogue of manuscripts at Munich, where the example is dated 1294 – , and, second, if we browse through the surrounding images, relating to analogous Greek crosses, most of them relate to the 1290s and none are for the years 1283 or 1284.
Returning to entry n. 5410, what was the document involved? Unfortunately, and unfortunately is something of a misnomer, the reference Briquet provides to the archive series is a blithely unhelpful “Podestà” (i.e. Mayor) for Bologna’s State Archive, located just by the city’s Piazza Maggiore and very pleasant place to work (Bologna’s fame as a gastronomical paradise is entirely deserved!). One wins no prizes for guessing that “Podestà” is a sizeable archive in its own right. One would like to think that subsequent scholars, given that a century and more has passed, have sought out this earliest example recorded by Briquet. Like Hell! The whole issue remained as dead as a doornail, up to very recently, when Bologna scholar, Nicoloangelo Scianna, deserving of high praise, began a systematic trawl through the late Thirteenth-century documents in the Podestà series. His study, published in 2009, sadly – at least for the moment – has had to admit defeat, since it failed to uncover a watermark with the sought-for date and matching Briquet’s tracing.
No one is accusing Briquet of fabricating evidence or anything similarly heinous. He definitely saw something with that date and, even if it cannot be found today, it is important to understand what it was and why he himself expressed a doubt. My private suspicion is that the sheet was a wrapper, i.e. the outermost in a gathering, devoid of writing and added to protect sheets on unwatermarked paper with the date 1282. In my own limited experience of the Thirteenth-century material in the Bologna archive, I have noticed several instances of this practice, which is confirmed also by Scianna (see below). Inspection of the original tracing in the Briquet archive at Geneva sheds no further light on the matter and likewise Briquet’s diary for his first Italian journey in 1889-90, when he visited Bologna makes no reference to the discovery of a watermark, dubitatively, dated 1282 (Papiers Briquet, n. 40). Whatever Briquet traced and dated with an element of uncertainty, therefore, has either been lost or still has to be found (it is possible that in the century and more between us and Briquet that the document has been transferred to another series). Of course the great Swiss traveller deserves a slap on the wrist for not giving a more precise reference to the document, but in his defence it is possible that he did not realise its importance when he found it. His exploration of the Bologna archive occurred at an early stage in his vast project and it is likely that he did not as yet have the full picture of the chronology of early watermarks nor understand the significance of that particular find.
As matters stand, therefore, the most reasonable date for the appearance of the first watermarks becomes the second half of the 1280s, at least as far as reliably dated documents are concerned. Scianna’s study did uncover examples of the Greek cross watermark in Bologna documents with slightly later dates. However, these same dates are suspect. One could be dated 1284 and another 1285, but in both cases the sheet of paper was not written on at the time, but functioned as a wrapper for other leaves, so it could have been added at a subsequent moment. Briquet assigns a tentative 1285 to an eight-petalled flower (n. 6584; not confirmed by Scianna, who does however confirm the other Briquet date of 1294) and also to a fleur-de-lis (n. 6710, again not confirmed by Scianna, who does confirm the other Briquet date of 1293). In both cases, even without locating the items concerned in the Bologna archive, the advanced shape of the watermark suggests that the documents thereon were recopied at a later date. For 1286 Briquet found an example of a letter I (n. 8247, not mentioned by Scianna), while for 1287 in the “filigranes indeterminés” he traced a pear-shaped object, which Stevenson in his corrections suggests is a buckle or fastener (n. 16005). The latter is the first example outside the archive at Bologna, since it was discovered in an archive at Torcello in the Venice lagoon. (What is really required is a more systematic exploration of the unwatermarked sheets of paper in the Bologna archive belonging to the period 1250 ca. up to 1286, since much of it is clearly made with Western rather than Oriental or Arab techniques. To some extent the question of the earliest watermarks has distracted scholars from much more important issues.)
Some confirmation for the mid-1280s for the first appearance of watermarks comes from another pioneering repertory, which most scholars have missed, Luigi Volpicella’s 1911 listing of watermarked papers up to 1500 in the State archive at Lucca [6e. Tuscany]. He reproduces two watermarks, both a giglio or lily, in documents dated 1284 (n. 1) and 1286 (n. 2), and also indicates the archive source as Estimo, n. 7. Obviously this is something that could be followed up with profit.
As a digression, and also as an example of willfully wrong-headed analysis of watermark evidence, I mention that in 1953 a palaeographer and local scholar from Cremona, Uberto Meroni, claimed a much earlier example of a watermark in a document dated 1271 . The manuscript concerned was a composite assembly of documents relating to the Benedictine monastery of San Pietro al Po, near Cremona, formed of some 18 separate gatherings. The watermark is found in the first three leaves, which on two pages contain records dated 1271, but elsewhere contain records dated 1365. The watermark (reproduced in the article) is a letter “F”, similar to several traced by Briquet (nn. 8145-48) in documents dated between 1314 and 1335. Is it really necessary to shout from the roof tops, yet again, that the date of a document written on a piece of paper may have nothing to do with when that piece of paper was made? Not only in this case is the make-up of the document highly suspicious, but the mixture of dates in the same leaves should have induced the strongest of doubts. Rather than argue that watermarks appeared in a fairly sophisticated form as early as 1271 and then disappeared for nearly two decades or, in the case of this particular design, for over forty years, it is much easier to presume that the notes relating to 1271 were recopied at a somewhat later date. Nevertheless the date “1271” has been cited in other studies, such as the small catalogue accompanying the Bernstein bull's head watermark exhibition , without questioning the sheer implausibility of the evidence (and this is the sort of thing that gives paper studies a bad name).
Watermarks. Names and Shapes, Ups and Downs, Lefts and Rights
Early references to watermarks in Latin define it with the Latin term signum. Lists and inventories of Medieval and Renaissance paper merchants also make reference to named watermarks, as a way of distinguishing different sorts of paper, but unfortunately without providing a key for subsequent interpretations.
Twin moulds, and thus twin watermarks, evolved naturally from the practice of having two men, the vatman and the coucher, working together as a team. At some point the practice was established of organising twin moulds as if they were a pair of shoes, with one watermark in the left-hand mould and its twin in the right-hand mould. Historical watermark scholarship has, however, been remarkable for the total disinterest it has shown for this phenomenon, so no solid documentation has been accumulated as to when and where this practice first evolved. One fact about the presence of twin watermarks in any supply of paper, that has troubled some scholars, or made them reluctant to stick their necks out on the matter (for example, David Woodward), is that there is no surefire proof. The twin moulds concerned have not survived; all that is left is the paper made on those moulds. The limits of inference in bibliography have often been discussed and there is no room here for a “Papermakers of the mind” conversation. It is largely a matter of common sense. If a document displays a pair of look-alike watermarks, alternately in the right/left halves of their respective moulds, in a reasonable quantity, without other marks getting into the mix, the most sensible and economical assumption is that these are twins. It remains an inference; paper studies, however, are not nuclear physics and the best we can ever do is inference, albeit an inference based on a high order of probability. But the heart of the whole watermark discussion is here. Twin moulds, twin watermarks, the twin little girls in The Shining by Stanley Kubrick (1980). Twinship establishes identity and a paper-stock is only identified when both its watermarks are clearly recognised.
How watermarks were shaped from the Middle Ages to the Renaissance is largely a matter of guesswork, since no physical examples have survived; we have only the indentations they have left in countless sheets of paper. The watermarks obviously had to be as similar as possible, since they were marking alternate sheets in the same post, but how similar varies according to the period and the area. If we take the trouble to identify twin watermarks in Medieval rather than Renaissance sheets of paper (taking the Middle Ages conventionally as lasting at the most up to half-way through the Fifteenth century), in many cases they are easily distinguished, not just in terms of their placing on the mould, but also in terms of size and shape. It is plausible therefore that the mould-maker allowed himself a certain freedom in bending and shaping the wire. A sort of progression towards uniformity, linked to the increasing complexity of the figures, was inevitable, so that watermarks were increasingly shaped around patterns and thus are ever more difficult to tell apart. Patterns could take the shape of nails in a board or of a form cut into a wooden mould into which the heated wire was forced. Their use was encouraged by the increasing complexity of watermark structures and so by the Sixteenth century the practice seems to have been more or less uniform, requiring an expert eye to distinguish the twins in a pair, unless of course the scholar is versed in the not particularly arcane skill of telling mould/felt side apart, so that the identification recognises whether the mark is in the left/right hand side of the mould . Of course the same pattern could make dozens of virtually identical watermarks and also be kept for an indefinite period of time. Nevertheless an awareness of the pattern as a ‘further level of complexity’ does not seem to have impinged on the thinking of most watermark scholars. And this is unfortunate.
An important structural change in many Fabriano moulds (and related centres) consisted in putting the watermark on a supplementary chain-line, which in turn is placed between two more widely-set chain-lines, i.e. if the average distance between chain-lines is about 30 mm, the distance between the two chain-lines will be about 55 mm, with the supplementary chain-line in the middle, at a distance of 27 mm on either side. Though no examples of such early moulds survive, it is unlikely that the supplementary chain-line had a supporting rib beneath it, though it may well have been held in position by a thick wire, as with a tranchefile. This feature is also distinguished by the fact that it did not place the watermark in the centre of one half of the sheet, but shifted approximately a chain-line towards the true centre. The vagaries of observation, since repertories rarely, if ever, specify the presence of this supplementary chain-line, mean that it is not easy to date this innovation, but it seems to emerge in the last quarter of the Fifteenth century.
One small point needs to be made for descriptive purposes. Over 95% of watermarks have a clearly recognisable up and a clearly recognisable down. In other words there is an automatic consensus that humanoid figures (angels, mermaids, pilgrims, etc.), animals (dragons, oxen, unicorns, etc.) and most inanimate objects (anchors, bells, fruit, hats, etc.) should be represented with the top at the top and the bottom at the bottom (is that too complicated?). There are a few exceptions: the flower bloom watermark, essentially a circle with a surround of petals, common in Milanese incunabula, is possibly the most widespread and is a considerable nuisance in identification terms, especially when one is trying to recognise a pair of twin watermarks. It had a run of well over a century, see the examples at and surrounding Briquet n. 6600, also nn. 6621-22. Likewise the pommelled cross in a circle, typical of Colle Val d’Elsa in Tuscany and beloved of Roberto Ridolfi, does not have a discernible up or down . Yet another example that has brought complaints from filigranologists is the Tudor rose, employed by Britain's first papermaker, John Tate. The simple fact of having an up/down makes the comparison and description of watermarks enormously simpler, though the reason was probably practical. In the paper mill the deckle had to fit both moulds loosely but precisely, and so it was better to put it on always in the same way. Having a clear direction on the watermark visible in the centre of one half of the mould naturally made this simpler. As a corollary it should be added that the watermark is always aligned with the long side of the forme, something that can be useful when trying to identify a complicated format.
Along the same lines, when we start talking about the placing of watermarks, and thus identifying the individual watermarks in a pair of formes, by the by and not unsurprisingly, scholars have been anything but consistent and coordinated in their language. The most belaboured, and contentious, issue has been to decide which side of the sheet the watermark should be described from, which is surprising given that there are only two possibilities. The overall preference – favoured by scholars such as Allan Stevenson, G. Thomas Tanselle, Alan Tyson, and Paul Needham – has been to view the sheet of paper “mould-side upwards”, i.e. the reverse of the layout of the mould, since it is the more convenient for actually seeing, tracing or rubbing, or photographing the watermark. Papermakers were well aware of this fact and therefore when a mould maker inserted letters or a name as part of the watermark, the general, if not invariant practice, was to shape it in mirror writing, so that to read the text the sheet is necessarily viewed from the mould side. A minority, including myself, have felt that it makes better sense to describe the watermark as if the original mould were in front of us, which entails viewing, or at least describing, the sheet “felt-side upwards”, with the inconvenience of tracing words or letters in an Alice-through-the-Looking-Glass fashion (but it is easy enough to transcribe the text and add the fact that on the mould it appears in mirror writing). In its quadrilingual standard issued in 2013 the IPH compounds the confusion by advocating the “wire side facing down” (paragraph 3.0.17) in its English, German, and Italian texts, i.e. looking at the sheet from the felt side, and the opposite in the French one (at least this is how I interpret the phrase: “Il est recommandé de recueillir les données sur la face inférieure (c’est à dire la face de la feuille qui a été en contact avec la forme durant sa formation)”). Common sense suggests, however, that there is no point in a Swiftian conflict between Big enders and Little enders, since watermark images were conceived and made as reversible, and therefore dictatorial absolutes have no place in their study. Individual scholars ought to apply whichever method suits them best, but simultaneously be very clear about whichever system they are using.
The other related, and likewise important, question has been how to name and distinguish the twin watermarks, sometimes in quite long runs over more than one gathering, especially when mapping out their distribution in a document, usually a manuscript. Three different solutions have been experimented in the discussion about the “1460” Mainz Catholicon, in which brilliant pioneering scholarship identified watermark twins in a variety of states . For the sake of clarity, it is easiest to deal with them in reverse order. The formula proposed by Needham (1982) and practised by him in subsequent writings is to describe a watermark placed in the right-hand half of the original mould as “Mould-side Left” (mL), i.e. as it appears in the sheet viewed from the mould side, and one in the left-hand correspondingly as “Mould-side Right” (mR) . This language is self-explanatory, though Needham, as a specialist in printed documents, has not approached the issue of how to map the watermarks an extended document such as a Medieval manuscript. Eva Ziesche and Dierk Schnitger (1980) on the other hand prefer to view the sheet from the felt side, i.e. as the watermark appeared on the mould, labelling a watermark in the left half as “a” and that in the right half as “b” .
A rather more elaborate nomenclature, applied initially to his study of the Catholicon (1973), was excogitated by Theo Gerardy . He also employed it subsequently in his book-length study of the watermarks in the archive at Fribourg (the one in Switzerland, or Freiburg im Üechtland, see [6k]), in order to describe twin watermarks in documents from the first half of the Fifteenth century (1980). The system envisages a sheet of paper, folded as a folio, with the watermark the right way up, in which the mould-side falls on the verso of the first leaf (therefore with the watermark positioned in the right-hand side of the original mould, or what Needham calls mL and Ziesche-Schnitger call b). In this instance the watermark is designated zugewandt, i.e. “turned towards”, or, more simply, “Z”. In a sheet from the opposite twin mould the watermark, seen in the same fashion, necessarily falls on the recto of the first leaf (therefore the watermark was positioned in the left-hand side of the original mould, or what Needham calls mL and Ziesche-Schnitger call a), and is designated abgewandt, i.e. “turned away”, or, equally simply, “A”. In description, therefore, the conjugate leaves that make up the same sheet, in a folio arrangement, are defined as nZ/-Z and nA/-A, where “n” represents the watermark and “-” its absence. This system has also been publicly espoused by Scottish scholar, Roderick J. Lyall, who adds a further twist to the cocktail by indicating sheets in which the watermark is upside-down (as all too frequently happens) with an asterisk . In a hypothetical example provided by Lyall, involving a regularly quired folio gathering in sixteen, with the sewing at the centre between ff. 8 and 9, the resulting description appears in Figure 1.
Figure 1. The distribution of watermarks in a hypothetical folio gathering described by Lyall.
This seemingly intricate pattern translates as follows: the said gathering has the expected eight watermarks: six from the mould in which the watermark was in the left-hand half (mR), one the right way up at f. 15, and five upside-down at ff. 4, 5, 6, 7, and 16; two from the sister mould, in which the watermark was in the right-hand half (mL), both upside-down, at ff. 9 and 14. (It is possibly not a very helpful hypothetical example, since six watermarks are from one twin, two from the other, and only one watermark is the right way up!) Once the system has been learnt, it is not difficult to interpret, but remains rather hard on the eye.
On the whole the Gerardy-Lyall annotation seems an over-complicated way of describing something simple, and perhaps hinders the real aim of the exercise, which is to identify the twin watermarks, tell us where to find them, and show their numerical differentiation. It is furthermore a system that might struggle with extremely large gatherings (in Medieval archive documents up to fifty leaves in the same gathering are not unusual). In my view, if the gathering is regularly quired, it is superfluous to say that there is no watermark in the other half of the sheet and even the indication that it is upside-down appears an excess of zeal. Applying instead the method I currently employ to map twin watermarks in the Medieval archive of Udine, of which more in the following chapter, the diagram of Lyall’s hypothetical gathering appears in Figure 2, in which the marks are distinguished as R [right] and L [left] on the basis of their position in the original mould (for the sake of argument I maintain the asterisk, though in my own work I do not employ a symbol for upside-down). The conjugate leaves forming the gatherings are listed in opposite outer columns, with the watermarks in the inner columns related to the relevant conjugate.
Figure 2. The distribution of watermarks in the same gathering with the system practised by Neil Harris.
The principal objection to this layout is that, in a published form, it is wasteful in terms of space and therefore a more compact solution would have to be worked out. On a webpage, however, or just as a research tool, it has the advantage that information can be taken in at a glance, it is infinitely extendable and thus can accommodate very large gatherings, and the identification of the twin watermarks is immediate and self-explanatory. Likewise, instead of L and R, the system can easily accommodate the mR and mL suggested by Needham.
A further, and potentially unwelcome, complication to any annotation for Medieval watermarks is that well into the Fifteenth century it is quite common to find both marks in a pair placed in the same half of their respective moulds. Since Ziesche/Schnitger and Needham all deal with late Fifteenth-century printed documents, the problem does not pose itself. In his study of the Freibourg archive, covering the years 1402 to 1456, Gerardy employs the locution A I and A II, or Z I and Z II, which is perhaps cumbersome. In the tables written for my ongoing project on the Medieval archive in Udine I simply double the letter for the second watermark to appear, i.e. L and LL, and R and RR.
In descriptive terms, watermarks always subsist in a looking-glass world, so that the seemingly innocuous terms left and right are potentially dangerous and misleading. Although some sort of general labeling is indispensable, derived from the position in the original mould, when talking about details, it is better on the whole to employ unambiguous terms such as inner (i.e. towards the centre of the sheet or mould) and outer (i.e. further from the centre of the sheet or mould). This solution does oblige the user to be constantly aware of where the edge of the sheet is, but is quite workable.
Countermarks, Cornermarks, and Other Extras
In the latter half of the Fifteenth century papermakers around Lake Garda in Italy, in particular in the “Valle delle cartiere” above Toscolano, who in other words were supplying the burgeoning Venetian printing industry, introduced the “countermark” . If we imagine looking at a mould, in which the watermark is in or close to the centre of the right-hand half, the countermark is situated in the lower left-hand corner. (Rather than “countermark”, Paul Needham employs the useful term “cornermark” to describe these early examples; this seems an improvement on the “edgemark” favoured by Allan Stevenson). This countermark is never large, usually about 40×30 mm, frequently smaller, taking the form of a couple of letters, for instance A-B or Z-A, united by a cross or a leaf-symbol. The rule about the position of the countermark in the corner opposite to the main watermarks seems almost invariable, but I have encountered an exception in the final volume of the 1498 Aldine Aristotle, where an upside-down ‘A’ is found in some sheets in the corner above the anchor watermark. It is probable, however, that this is a simple placing mistake.
One rather unusual paper-size, something classified by Paul Needham as “Half-median” [see the previous Chapter], in which the original sheet measured approx. 255×360 mm, seems to have anticipated the countermark by placing the watermark, a small six-petalled flower, in the corner of the mould. This paper supply is used, for instance, in the already-mentioned mixed format Biblia latina printed in Venice by Franz Renner in 1480 (ISTC ib00566000), where the effective difficulty in understanding the placing of the watermark, especially when the copy has been cut down, has in the past led authoritative catalogues to believe that the sheets concerned were quarters of Royal (GW 4241, IGI 1661). Beginning with Pellechet and Proctor, however, other repertories recognized that the real format of these sheets was folio. In another unusual solution, found in the Super-median sized sheets of the Latin Bible printed in Venice by Paganino de’ Paganini in 1495 (ISTC ib00597000), two small five-petalled watermarks are placed in diagonally opposite corners of the moulds. More data needs to be gathered from Fifteenth-century paper supplies, but it is plausible that these positional experiments were a way of signaling a special sheet size.
The purpose of these early countermarks was obviously to identify the provenance of the sheets of paper: in a few cases, albeit not many, it has been possible to match the initials with a particular family of papermakers. But the system was not intended to be decipherable for the general public; it responded rather to the need of the paper-merchant, who was often a wealthy businessman living in a town close to the papermaking district, while the papermills were spread out over the hills in the surrounding countryside. Since many of these mills were making sheets of paper with the same watermark, something else was obviously required in order to allow the owner, especially if there were complaints about quality, to recognise which mill a particular item came from. Placing a couple of letters in the corner of the sheet meant that the trace left in the paper was easy to see and could be read without going to the trouble of lifting the whole sheet up to the light.
Among the great merits of Briquet is that he is very much alive to the importance of the countermark and is scrupulous about recording its presence (see, for instance, his entries for Ancre). He gives the earliest example he knows as 1483, without however telling us where (I, p. 14). I have searched through Les filigranes, including its electronic version, without finding the example of the same, which he may therefore not have included. A 2004 discussion of countermarks by Paola F. Munafò and Maria Speranza Storace, however, confirms 1483 as the earliest recorded example found in the Latin Bible printed in Venice by Johannes Herbort (ISTC ib00579000).
The practice of countermarking in the corner of the sheet in the Toscolano district lasted well into the Seventeenth century, but, despite its undoubted utility, did not spread to other areas of Italy or to other countries. For bibliographical purposes countermarks of this sort are an absolute boon, given the right bibliographer. First, since letters of the alphabet have a clear indisputable direction, they make it very easy, as Conor Fahy has pointed out , to tell twin moulds apart, i.e. if the letters are A-B, in a mould in which the countermark is in mirror writing in the bottom left-hand corner, the B will be the outer letter; in the twin mould, in which the countermark is in mirror writing in the bottom right-hand letter, the A will be the outer letter. Second, in large and medium-sized formats, otherwise known as folio and quarto, the countermark usefully tells us that the other half of the sheet, the one without the official watermark, still comes from the same mould, or conversely serves to identify cancellantia introduced with less than a full sheet. Third, in small formats, such as octavo and less, for printed books it tells us the imposition of the typographical forme, i.e. if the countermark is in one of the first four leaves of an eight-leaf gathering, the scheme is common or centripetal octavo; if it is in the last four, it is inverted or centrifugal octavo. Fourth, and not least, the countermark generally falls outside the part of the leaf obscured by print, and never ends up in the inner margin, so it is often more visible than the watermark true and proper (there is however the danger of the binder's plough, which sometimes removes it in part or in whole, so it is important to pay attention to the size of the copy).
In France in the later Sixteenth century legislation required the paper-maker to include the family name in the watermark: so French paper of this period generally contains a sort of scroll underneath the symbol with a name. In Italy much the same process occurred spontaneously through the countermark, which, beginning in the last quarter of the Sixteenth century, began to shift to the centre of the half of the mould opposite to the main mark and became correspondingly much larger. The earliest example I have encountered is a large “TC” with the watermark of an angel in the opposite half of the sheet in the Historiarum de regno Italiae by Carlo Sigonio, published at Bologna in 1580, but I am sure that systematic research would uncover earlier examples. Again watermark repertories are not as helpful as they might be in documenting the evolution of this process, since the best of them (Briquet) stops in 1600, when the process is only just getting under way, while the others cannot be trusted to provide reliable information in the matter of countermarks. My rather provisory impression is that the consolidation of the countermark, placed opposite the mark in the other half of the sheet, happened in the late 1640s. In some cases the countermark was accompanied also by a cornermark identifying the papermaker.
In the Seventeenth century the initials of the countermark were increasingly transformed into a full family name, which rather curiously took the mark back to its original function as an indication of the making or the provenance. By the Eighteenth century these signs are highly evolved and appear in symmetry with the watermark in the other half of the sheet. Examples are Miliani in Fabriano, Magnani in Pescia or Villa Basilica, and Whatman in England; at times the emphasis falls rather on the name of the mill, such as La Massa at Villa Basilica near Lucca, La Briglia on the Arno above Prato, or the Whatman Turkey Mill in Kent.
From the Eighteenth century to the early Nineteenth the practice of including the date as part of or as the entire countermark became increasingly common. At times this was obviously a marketing ploy, aimed at showing that the paper was “new”, but these chronological pointers should always be treated with caution, since the seemingly helpful indications can bite the unwary and the ill-informed. Although they establish an unquestionable terminus ante quem non, which can nevertheless prove valuable where the date of a document is highly uncertain, at the other extreme, that of establishing a more hypothetical terminus post quem non, they can be deceiving. In the first place a mould with a date could remain in use for a longish period of time: it is known that the Whatman mills did take the trouble to update their moulds at the beginning of each calendar year, but they were an exception rather than a rule. In the second place the inclusion of the date sometimes had a different purpose, for instance to benefit from a lower rate of tax, so that mould-makers continued to put that particular year even in later periods: for this reason the date “1742” in French paper and “1794” in English paper should always be treated as suspicious and as having little or nothing to do with the date in which the sheet was really made .
By the late Eighteenth century moulds have generally acquired quite a lot of writing and numbers that make watermarks complex and easily recognizable, while the scripts also ensure that it is easy to distinguish the felt/mould sides of the sheet. Unfortunately, however, changes are afoot. Italian papermakers generally stick to the long-standing right/left principle with the texts on the moulds in mirror writing. In Austria, on the other hand, as Alan Tyson shows in his brilliant study of Mozart’s manuscripts (and by “brilliant” I mean “absolutely brilliant”) , mouldmakers put scripts in mirror writing in only one of the twin formes, so that the sheets are ostensibly much more alike, with the watermark and countermark always in the same relationship, as far as the eye of the casual observer is concerned. The sameness of course means viewing one sheet from the mould side and the twin sheet from the felt side. Until more thorough research is done on twin moulds and on the positioning of watermarks in the moulds, it is impossible to say how widespread the phenomenon was.
As a final addendum, or complication, and invitation to keep looking at paper with a critical eye, another feature I have observed in Italian paper of the late Eighteenth century in Tuscany is the inclusion of yet another watermark in order to number the pair of moulds. It takes the shape of a single digit – 1, 2, or 3 – placed in correspondence with the bottom of the central fold of the finished sheet, and therefore it can only be seen in unbound or disbound sheets. Its function was very obviously to distinguish pairs of moulds, especially in large mills with more than one vat, that were otherwise identical. The same number was placed on both the moulds in a pair.
Apart from purely visual methods, numerous studies provide examples of descriptions of paper and watermarks, with very diverse levels of competence and ability. In many instances seemingly elaborate and exhaustive descriptions are furnished by scholars, who have nevertheless failed to understand that watermarks are twins and thus waste time, both their own and – more’s the pity – other people’s. There are also plenty of instances in which the same (pair of) watermarks have been counted twice over, since they have been viewed both from the mould and the felt sides of the sheet.
In my humble opinion, the only genuinely effective way to describe a watermark is to procure a quality image. Nevertheless even the best picture should be backed up with written notes, drawing attention to significant features, since even the best image rarely shows all the characteristics of a watermark, while there are often circumstances in which, for reasons of cost or of practicality (for instance, a small format or a tightly bound book), in which it is not possible to get a picture and therefore a scholar has to resort to a written description. Attempts have therefore been made to prescribe uniform solutions in the description of paper and of watermarks, albeit with mixed results.
So, without being prescriptive, let’s make some suggestions.
As stated above, it is always a good idea to start by defining the purpose of the description.
If the aim is to label a certain supply of paper, especially in one or more editions of a printed book, where further copies can always be added to the equation, there is sometimes little purpose in going to all the bother of distinguishing twin-watermarks, right and left-hand side placings, and so on and so forth (though one should always have these more refined methods available in one’s personal arsenal). In these situations it is nonetheless important to know about the sheet-size (or at least provide the measurements of the largest copy encountered), the type of watermark (noting any unusual features, such as a cross or a flower above), the distances between chain-lines and the density of the wire-lines, and eventual other characteristics, such as the presence of countermarks, supplementary chainlines, or tranchefiles.
If, however, it is a matter of identifying the watermarks comprising the twins in a pair, it is important to begin by clearly distinguishing the left- or right-hand moulds (alternatively one can label the sheets as mirror-image right- or left-hand indentations when the sheet is examined from the mould side). The next stage is to reconstruct the overall lay-out of the two moulds, i.e. by working out the original sheet-size and calculating the number of chain-lines and the intervals at which they appear. It is also a good idea to measure the average number of wire-lines in a distance of 20 mm. In describing the watermark, apart from its measurements and characteristics, it can be helpful – as Tanselle advocates – to provide indications of the distances between it and the chain-lines. If the final intent is to insert the description in a database, it is possible to include the codes and criteria espoused by the IPH in the Bernstein project (see below), though these are not always straightforward to employ or to remember.
It should always be remembered however that, as far as recognising any individual pair of marks goes, even a very elaborate description is only a palliative. The only really effective record is the knowledge and understanding applied by whoever is studying a particular set of marks. Watermark scholars are like shepherds, who know every single sheep in the flock.
A word of warning. It happens quite often that in a book or document the inquiry, even if it is conducted properly, finds that there is no way of discriminating the pairs of watermarks. Just to give an example, the edition of the De Cardinalatu by Paolo Cortesi published at San Gimignano in 1510 employs, among others, four watermarks containing a fleur-de-lys. They are easily told apart: two are right-hand and two are left-hand; which however goes with which? On the evidence of this one book there is no way of telling. In the long run, but it would take a certain amount of time and trouble, it might be possible to find another book of the same period, in which the paper came from only one of the two sets involved here and so clearly distinguish the pairs. What specialists often seem to fight shy of, however, is the embarrassment of admitting publicly that they do not know and so they tend to skate over the issue, by blithely omitting any mention of the twin relationship. This tacit untruth in reality helps no one. In a situation such as this, it is much better to admit that one does not know. A graceful admission of impotence is much better than a bumbling attempt to disguise incompetence.
As in the well-known instance of the Fuolornis Fire Dragons in Brequinda in the Foth of Avalars (another reference to that book!), what purposes to be a guide tries to keep up to date (“purpose” being the operative word). So, rather than write about reproduction techniques as if they were extant and freely available on the market, I have sought to do something more practical, and even more sensible. In other words to assess, through contacts with the imaging services of major institutions that have produced β-radiographs and analogous material in the past, the actual status quo (in 2016, but things evolve rapidly, especially where obsolescence is concerned).
The earliest attempts to reproduce watermarks were freehand drawings, which were printed either by cutting the images onto woodblocks or engraving them onto copper plates: the results therefore give only a very approximate idea of the original design. They often excluded any indication of the measurements of the watermark, nor did they provide information about the placing with reference to wire and chain-lines.
The subsequent and most widespread method, still used today, is the tracing, which is economical, fast, easy to reproduce, and thus was the staple procedure of pioneering scholars such as the elder Zonghi, Briquet and Piccard. In his travels Briquet carried with him packets of tracing paper, previously cut up into small rectangles of the appropriate size. Here is his description of how he went about his task: “...Lorsque, et c’est le plus fréquent, on doit relever la marque dans un volume relié, on est obligé d’abord de chercher le feuillet où cette marque apparaît de la façon la plus distincte, puis de placer le volume sur une table, près d’une fenêtre, en pleine lumière, à l’hauteur d’oeil et de soutenir le feuillet choisi par une plaque de verre de dimension convenable. On peut alors calquer commodément et exactement. Il sera bon de dessiner, en même temps que le filigrane, les pontuseaux entre lesquels il est placé et de noter les fils de la vergeure pour juger de leur écartement.
Translation: When, as is most often the case, one has to trace the watermark in a bound volume, one first has to find the leaf where the said watermark is at its most visible, next put the volume on a table, near a window, in full light, at eye level, and hold the leaf concerned up with a piece of glass of an appropriate size. In this way one can make a tracing with comfort and precision. Together with the watermark, it is a good idea to trace the chainlines on either side and to note the wirelines in order to show the distance between them (Les filigranes, cit., I, pp. xvii-xviii).
Quality tracings require a lot of skill and experience, since it is far less easy than it seems, as classroom experiences have duly shown, to reproduce accurately a watermark. Tracings have the unquestionable advantage of being the same size as the original (though some modern reprints of Briquet – the worst example is the 1985 New York edition – have reduced the designs by as much as 20%: I hope that in the special fiery Hell reserved for publishers these people are in for some particular noxious form of eternal perdition). If they are the work of a genuinely scrupulous scholar, such as Briquet (funny how that name keeps coming to the fore), who also records the position of the watermark with respect to wire and chain-lines, not to mention giving the sheet-size, tracings can be extremely useful. On the other hand, the act of tracing a watermark through another sheet of paper reduces visibility, so that even the best scholars rarely record minor details, such as sewing dots, unless they are extremely obvious. In the final count, therefore, tracings are never going to be exact enough to allow us to be certain that we are dealing with the same (pair of) watermarks.
Where the original sheet of paper is reasonably dense and where the surface has not been disturbed by printing, as an alternative to tracing, an accurate image can be obtained through a rubbing (frottis). The principle is identical to that employed to reproduce the blind tooling on bindings or memorial brasses in churches. After identifying the mould side of the sheet, take a piece of thin paper, put it on the leaf (it is best to have a weight, or a helper to hold it in position), and shade over the surface with a very soft pencil. The results are useful for the purposes of private study, rather than publication, but the method is used regularly by Paul Needham who has reproduced some examples (see, for instance, his article in Puzzles in Paper). Further examples are viewable as digital images in the large collection of images now on the Wasserzeichen des Mittelalters interface of the Austrian Academy for Sciences, lead partner in the Bernstein project, and in the material relating to Dutch incunabula collected by Gerard Van Thienen on the website of the Koninklije Bibliotheek in The Hague (both accessible also through the Bernstein website) . Although the system is less arbitrary than tracing, some detail is lost, and of course (this is the real down side) relatively few libraries or archives are willing to allow scholars to do this to the documents in their possession.
Over the course of the last hundred years photographs of watermarks have been published with a certain frequency and Briquet, for instance, includes a selection of examples in his preface, to which Stevenson adds others in his introduction to the Jubilee edition. The big traditional obstacle has always been cost, to which has to be added the fact that it requires a clever photographer, especially in illuminating the sheet, to get a decent result. Another major handicap is that where the watermark is obscured by handwriting or, worse, print, its visibility is seriously compromised: examples tend to be taken from occasional blank sheets or cases in which the watermark coincides with a gap in the writing. Further limits are imposed by the fact that photography rarely overcomes the obstacle posed by medium and small formats, where the watermark falls in the margin or is broken up into different leaves. (One small mystery, for instance, is posed by the truly excellent photographs displayed in Ridolfi’s 1957 pamphlet, which showed the watermarks in some quarto-format Florentine incunabula, where the watermarks fall in the inner margins. How did he do this? Quite simple, they were his own personal copies and he had them disbound for the purpose! It is nice to be a Florentine Marquis and have one’s own collection of rare Fifteenth-century books.)
The only paper-published work I know in which all the watermarks in a census or collection are reproduced by photography is that by Jane Roberts for the drawings of Michelangelo (1988), sponsored at the time by Olivetti . In this case the fact that most of the items were separate leaves, as well as the circumstance that the crayon or other materials employed for the drawings hardly obscured the watermarks, meant that good results were obtained. On the other hand the images are not always sufficiently distinct (not all the photographers involved knew how to capture images of watermarks) and the repertory fails to indicate the sheet-size, whether the image is taken from the felt or the mould side, and the effective dimensions of the watermark itself.
In the 1950s, as a somewhat unexpected spin-off from the USSR’s nuclear programme, Russian researchers discovered a method, in which images of watermarks were recorded with β-radiography (basically a low-powered x-ray). The first article describing the technique in a Western language appeared in 1961 . The operation is relatively straightforward: the sheet or leaf of paper is sandwiched between a weak radioactive source and a photographic negative; where the paper is thinner, i.e. where the watermarks and the chain-lines have left indentations, the radioactive particles penetrate more easily and thus expose the celluloid more fully. The outcome is an image in which the watermark appears as a white shape on a dark ground (obviously the contrast can be reversed in publication, although this rarely happens). The advantage is that the radioactive articles are not affected by writing or printing and so it is possible to capture a clear image even when the watermark is ordinarily obscured by a heavy layer of ink; the disadvantage is that, quite apart from the problem of getting hold of a radioactive source, the procedure requires laboratory facilities and trained technicians, so that the few instances, in which a large number of images have been acquired and published, involve large institutions, such as the Louvre, which has conducted research on the watermarks in its collection of drawings (De la Chapelle) , and the Newberry Library, which has published work on its Italian Renaissance map collection (Woodward) . For the revised version of this text an inquiry has been conducted about the availability of this technology and unfortunately it has proved obsolete just about everywhere. The original plastic plates containing the radioactive salts, often made in the 1960s, had a tendency to crack and deteriorate, while in more recent times the special negative paper needed to develop the image is no longer commercially available. Consultation with a number of laboratories, including the Bodleian and the Huntington, also established that the procedure had not been requested by researchers in the last decade, while safety concerns involving the radioactive “wafer” used to produce the β-rays have augmented. It should be noted, however, that major libraries often conserve archive negatives, so it might be worth checking whether the watermarks in a particular document have been acquired by β-radiography in the past.
An analogous procedure, or electron radiography, in which the same basic principle has been applied to obtain images of watermarks in Dutch incunabula, can be viewed in the WILC project hosted by the Koninklije Bibliotheek in The Hague, accessible also through the Bernstein project [23, 35]. In this case it was the achievement of a librarian, Gerard van Thienen, who managed to convince an important firm working in the field, the Röntgen Technische Dienst, Rotterdam, to construct a machine and to train technicians in order to conduct an in-depth survey of the watermarks in the library’s holdings of Dutch Fifteenth-century books. Inquiry to The Hague in 2016 did establish, however, that the collaboration with Röntgen ceased well over ten years ago, once the project had been completed, and that at the present moment the procedure is not available. The website of the project, which includes an excellent step-by-step demonstration of the making of electron radiographs, is not explicit about this fact, so be advised.
Again slightly different, in technical terms, but the same essential procedure, is the Soft x-ray radiography developed in the mid 1980s by Jan van Aken, formerly professor of dental radiology at the University of Utrecht. This method, which provides a high-quality result, has been applied to imaging watermarks in Dutch projects conducted by Theo and Frans Laurentius, as well as in the Bernstein project by Manfred Schreiner at the Academy of Fine Arts in Vienna. At the time of writing in 2017, this procedure is part of ongoing work and is still being used. Since x-ray procedures are demanded by archeologists and art-historians (for instance, to x-ray mummies or Renaissance paintings), who have infinitely more kudos and wealth than mere paper historians, the imaging services of major galleries and museums do have such services and can provide them on request. So ask (and if they say no, scream!!!).
A slightly different approach employs ultra-violet light (DYLUX method) and is associated above all with the name of Thomas Gravell . The technique is very similar to those described above: special photographic paper sensitive to UV light is placed behind the watermark and exposed to a fluorescent light source; the image of the watermark is not impeded by ink and so develops clearly. As a method, it is less effective than the x-ray procedures described above and, if the paper has been heavily written on or printed, the results can be poor. Its main application has been in the field of stamp collecting, where the paper is very fine and the watermarks often employ shading or other niceties.
Phosphorescence, infra-red imaging, and analogous methods, relating to the way light or heat pass through the sheet of paper, have been experimented with at different times, without however producing long-term solutions. These techniques give a result, which is inferior to the various x-ray procedures, but nevertheless require basic laboratory conditions.
An extremely interesting new development lies in the application of digital technologies originally conceived in the field of studying Medieval palimpsest manuscripts . Here the problem was to eliminate the upper, more recent, and heavier layer of ink, in order to reveal the earlier, fainter, often partially scraped text on the parchment. The technique consisted in identifying the chemical composition of the ink and removing certain elements of the spectrum, so that the upper layer vanished. It was discovered, almost by chance, that the same trick could be played with paper, although it is necessary first to photograph the sheet with a source of illumination behind it, usually a light-pad with optic fibres if the volume is bound. Subsequently, the ink is exported in a “virtual” fashion, leaving a clear image of the watermark. The method is effective also with printing ink and gives a final result, which aesthetically is more pleasing than β-radiographs. As matters stand, these procedures are jealously guarded by the firms that are developing them and in one tragic case have been lost on the death of their inventor . Nevertheless the future lies here.
So, to sum up, in 2017 the simplest, quickest, and cheapest way of imaging a watermark is to acquire a light-pad for backlighting and a good digital camera (or just a cellphone), and photograph it yourself, if and where allowed. The fundamental fact about watermarks is and remains that they are multiple images, so, if the source is a substantial manuscript or printed book in multiple copies, with luck, a blank leaf will pop up with a clear image. The big problem remains, however, those instances in which there are only a few watermarks, or perhaps only a single one, in the chosen source and it is obscured by handwriting or printing. At the present moment the staple methods of Twentieth-century watermark imaging have disappeared, overtaken by “obsolescence”, with the only partial exception of Soft x-ray radiography and the prospect of the conversion of radiography into a digital format; otherwise the hope has to be digital imaging, which has shown that it can be done, with excellent results (see Fotoscientifica), but at the moment has failed to provide a simple, easily affordable, widely accessible solution for those instances in which it is necessary to see through a burdensome layer of ink. The technology is there; it is just a matter of getting it to work. Here's hoping.
Nomenclatures and Classifications of Watermarks
These are something of a pain.
When my sweet little sister (who has now travelled gracefully past fifty) was just learning to speak, one of the first items in her vocabulary was the word “Hat”, which noun, once mastered, she delivered with great exuberance. One weekend during a jaunt out of town, on the way home the family stopped for refreshment. While we were still ensconced in the café, a large coach drew up outside, discharging a contingent of Women’s Institute members in full regalia; they entered in procession, heading for a eager cup of tea, and were greeted with the triumphal chant: “Hat!”, “Hat!”, “Hat!”, “Hat!”, “Hat!”, “Hat!”, “Hat!”, “Hat!”, for what seemed a very long time. The same fundamental problem is faced by any reader opening Briquet for the first time and turning to the entries classified as “chapeau”, which, beginning in 1309, roll on for many a page with only microscopic variation and nary a cup of tea (nn. 3353-3517).
How to call watermarks and what to call watermarks has always been among the chief torments of paper scholars. The problem is that, though these signs might have formed a coherent language for the paper-makers and the merchants of the time, for our purposes their meanings are often obscure. At Siena in 1334 a document mentions paper watermarked with the “signo della staffa” (crook or staff), while Zonghi cites the register of the paper-trader Lodovico d’Ambrogio at Fabriano who, between 1363 and 1366, notes the sales of some 58 types of paper, all recorded by their watermarks, a system to which the ledger provides no key . Likewise, in a famous document such as the Ripoli diary, or the notebook kept by the overseer of the press which worked in Florence from 1476 to 1484, acquisitions of paper often mention the watermarking: “due lisime di fogli comuni della colollna” (f. 5v: 1,000 sheets of chancery with a column watermark), “due lisme di fogli dagli ochiali” (f. 74v: 1,000 sheets with a spectacles watermark), “lisime due del segnio del guanto” (f. 78r: 1,000 sheets with the watermark of a glove), “due lisime di foglie da fabriano del segnio del balestro” (f. 83r: 1,000 sheets from Fabriano with a crossbow watermark), and “tre lisime di fogli comuni da colle del segno della croce” (f. 116r: 1,500 sheets of chancery from Colle Val d’Elsa with a cross watermark) . The mention of the symbol in these generally laconic records presumably had a meaning at the time, in terms of quality and price; unfortunately, we don’t know what. By the by, note that these early records all refer to the watermark as a “signum” in Latin or “segno” in Italian; the modern Italian term, “filigrana”, derives much more recently from French.
Out of these early dictions came the working vocabulary of the paper-trade. For our purposes matters are complicated by the overlay with sheet-sizes, which were not consistent in time or place, as well as the introduction of typologies with a specific purpose. The best-known example in English is “foolscap” (i.e. fool’s cap, a piece of headwear worn by a jester), born as a watermark, which became synonymous with medium quality writing paper (see examples in the OED).
Naming brings us to the real problem: arranging.
Repertories of watermarks, especially those aiming at giving a broad picture, always have the problem of how to display the images and thus present the information. Chronologies and geographical distributions for obvious reasons are untrustworthy; sheet-sizes can be misleading and often are not available. Experience of watermarks does teach that recognition is essentially image-based, so, whatever the system, it is best if it is simple. The most practical solution is therefore to confer a nomenclature and follow it. Of course there are problems, beginning with the language in which one does the naming. School-boy and school-girl (we must not discriminate) French takes us as far as “chapeau” in Briquet, but what is the word for “anvil”? (try “enclume”, nn. 5950-66, if you are really stuck). Or what about the strange object he denominates a “peson”? (a weight on a balance for the unenlightened). But, on the whole, for all its frenchifying, Briquet’s straight-forward alphabetical order works very well. It should be noted, however, that a project such as Watermarks in Incunabula printed in the Low Countries  skillfully circumvents the perils of franglais by employing the “English Typological Index” added by Stevenson to the 1968 edition of Briquet.
In recent years the IPH has promoted its own “Index of watermark classes and subclasses”, in six languages, included in its 2013 Standard, as well as in the 2002 book by Tschudin (pp. 291-353) . This employs 25 categories linked to a letter-code (there is no “I”), as follows:
A Human figures; men; parts of the body
E Fish; reptiles; insects; mollusks
F Mythical figures
G Plants (general); flowers; grass
H Trees; shrubs; creepers
J Sky; earth; water
K Buildings; parts of buildings
L Transport; vehicles
M Defence and arms
N Tools; equipment; clothing
O Musical instruments
Q Miscellaneous objects
R Insignia of rank; mace, sceptre, jewellery
S Religious or magic symbols and signs
T Heraldry; coat of arms; mason’s marks; merchant’s marks
U Geometrical figures
V Numbers; numerals
W Individual letters
X Monograms; abbreviations with letters
Y Names; words
Z Unclassifiable watermarks
Any and every classification theory denies that categories can be reserved for objects, things, or concepts that fit in nowhere else (and here there are two such: Q and Z), so some sort of reservation has to be expressed. Briquet, to my mind, puts things better, when in his pioneering article on the watermarks in the archives of Genoa (1888), he includes a category for ‘Signes dont le sens nous échappe’ (p. 114). Some of the identifications in the examples provided by Tschudin are also dubious: what he calls a “messenger, traveller” (p. 292) is identifiable by the staff as a “pilgrim” (as Briquet recognises and classes as “Homme. Pèlerin”, nn. 7563-7607), while his “acorn” (p. 317) is really an “oak-leaf”. One also wonders why the classical gods, such as Mercury and Neptune, are included in the human category (A) rather than in the “Mythical figures” (F), and so on. Just to return to one aspect of nomenclature, which has already been mentioned, if we follow Briquet’s labeling of a certain watermark as “main” (Fr.) or “hand” (Eng.), we classify it as category A, i.e. the human figure; if however we follow Renaissance sources, such as the Ripoli diary, and decide that it is really a “gant” or a “glove”, it shifts into category N, i.e. clothing. So all in all some guesswork is involved. The IPH system has been applied in some small-scale projects, most notably David Woodward’s 1996 book on the watermarks in Italian Renaissance maps , but otherwise has not gained wide-spread acceptance.
More recently, in 2012, the Bernstein project has published on its website an impressive glossary of ‘Watermark-Terms: Vocabulary for Watermark Description’, in seven languages: English, French, German, Italian, Russian, Spanish, and Hungarian (just Google), which is also being applied in a number of projects. Its most important application has been to the Wasserzeichen Informationssystem, which brings together various German institutions, beginning with the 90,000 odd tracings collected by Piccard and left by him to the State Archive in Stuttgart . The structure is very hierarchical: for instance, the principal motifs are: “Figures, anthropomorphic” (i.e. human beings), “fauna”, “fabulous creature”, “flora”, “mountains/luminaries” (don’t really see the connection), “artefacts”, “symbols/insignia”, “geometrical figures”, “coat of arms”, “marks”, “letters/digit”, “undefined mark”, and “work in progress”. So again, there are two categories for things we can’t really identify.
The discussion therefore remains open, but is unlikely to ever be satisfactorily resolved.