Palaeography: Analysis of the Letter-forms

From Alan Bowman and David Thomas, Vindolanda: the Latin writing tablets London: Society for the Promotion of Roman Studies, 1983, pp. 60-67

In the analysis which follows we have taken as a basis the careful analysis of letter-forms in ORC given by Mallon, which we have supplemented by reference to the more detailed study by Cencetti53. These analyses complement each other neatly since Mallon bases himself on first-century evidence, whereas Cencetti is primarily concerned with papyri of the second century. We have also found it useful to compare the examination of certain letters by Casamassima and Staraz54. The features which need to be borne in mind in analysing a script are the letter-forms, the angle of writing, the size of the letters and the ductus; by 'ductus' we understand the number of strokes which made up each letter and the order and direction in which each stroke was made55.

If the letter-forms of ORC have received a good deal of attention from palaeographers, the same is far from true of the ligatures in this script56. Indeed, the subject has been almost totally ignored in the recent studies discussed above, except in the work of Cencetti57. Fortunately, his analysis is careful, sound and very much to the point, so that it has been a great help to us in examining the ligatures in the Vindolanda tablets. Broadly speaking the ligatures in our tablets conform very closely to the pattern we should expect from the papyri as analysed by Cencetti.

By 'ligature' we mean the linking together of two or more strokes, which were originally separate, in a single movement of the pen or other writing implement, without removing the implement from the writing-surface. Such ligatures can be 'internal', within a letter58, or 'external', linking the last stroke of one letter with the first stroke of the next; the analysis which follows treats under 'ligatures' only the second of these types. Such ligaturing is, of course, the norm in present-day handwriting, but it came into use in the scripts of classical antiquity only gradually, and it is of some importance to see the extent to which it was practised in the Vindolanda tablets. A word of warning, however, needs to be added. It is often very difficult in any script from any period to be sure that a ligature has taken place, as distinct from instances in which two letters merely happen to touch one another because one ends at the point where the text begins. Such instances are of no palaeographical interest59; whereas examples of ligatures are always palaeographically interesting and are often of importance. It is usually possible to decide in studying a medieval manuscript whether a ligature has taken place or not by a minute exanimation of the original, sometimes taking into consideration a large number of possible instances. As we indicated above (Vol. I Ch. 4), such an examination of the originals of our tablets is rarely likely to be helpful; and as we are dealing with only very short texts, a particular ligature, or possible ligature, will usually occur only once, with no possiblity of a cross-check with other stances. In the remarks which follow we have tried to confine ourselves to ligatures of which we are confident (unless the contrary is indicated), but we are well aware that the photographs may on occasion have misled us into seeing ligatures where, in fact, none existed.

A. This is probably the most varied and interesting of all the letters. FIG. 11.1 shows the form in which it appears most often in the tablets (the form which is certainly the commonest in ORC generally). The main variants of this form are (a) the prolonging of Stroke 1 below the line, sometimes to a considerable extent (e.g. Nos. 2 (151), 4 (190), 30 (295)), a stroke which can end with a hook rising to the left, and (b) the appearance of Stroke 2 in a curved rather than a straight form (thus resembling r); this Stroke 2 can be written horizontally or even, in ligature, with a slope upwards to the right, e.g. suae, habeo (No. 37.8 and9 (225)). The part to the left of Stroke 1 is sometimes omitted (see No. 4 (190)), when the letter becomes hard to distinguish from i with serif (cf. No. 30.3 note (295)), or even c or p. The order of the strokes is not in doubt: ligatures to the left are always made with the top of Stroke 1, which must therefore have been made first and in a downward direction 60; see examples given under E, F, H, I, K, L, M, R, S, T, U; as this would suggest, such ligatures are very common. It is also common for the letter in this form to ligature to the right by means of the end of Stroke 2: e.g. ab (No. 30.6 (295)), ad (No. 4.9 (190)etc.; 30.5 (295), 37.11 (225)), am (No. 41.2 (323), 43.4 (325)), an (No. 13 (512)), ar (No. 38.2 (346)), as (No. 4.20 etc. (190)); note especially the ligature at in No. 4.41 (323), made by adding a link stroke, and the ligature hab in habeo and han in hanc (No. 37.9 and 11 (225)), where the whole letter a is made in a single stroke with a shallow loop to facilitate the ligatures61.

Fig. 11 Letter-forms in the tablets Image ownership: © Society for the Promotion of Roman Studies

Not infrequently a also appears in the form shown in FIG. 11.2, which corresponds to the second form in ORC in Mallon's analysis; this consists of three strokes, but the second and third may be made in a single movement of the pen62. In this form the letter never makes a ligature to the right. There is some reason to think, with Mallon, that this is an older form than that illustrated in FIG. 11.1. Examples of this form are to be found in Nos. 10 (224), 34 (218), 42 (324), and 48 (522), with an interesting variant in No. 43.5 (325), where the third stroke is made horizontally.

These forms are expected. There are, however, two forms of a in the tablets which we should not expect to find. (1) In No. 37.22 (225) the a in amicis is in the form illustrated in FIG. 11.363; in this form the letter looks much like one form of d. This is perhaps to be regarded as a half-way stage towards (2), the remarkable a to be seen in No. 45.2 (341), karisime: here Stroke 1 has completed its loop to Stroke 2 so that the letter looks very much like the a which was to appear in uncial64. The only a in contemporary cursive hands which is at all similar is, so far as we know, the one found occasionally in the pottery graffiti from Contodamagos65. Bookhands of this period sometimes have forms in three strokes in which the left-hand stroke combines with the cross-bar to form an angle66, and we think that the form in No. 45 (341) represents a rounding of this angle. Neither of these unexpected forms of a nor any of the usual forms considered above can be said to produce an a which can easily be thought of as having developed into the a of NRC. This is not surprising, since no form of a which obviously leads from ORC to NRC has yet been found in Latin papyri67.

B. On the forms of this letter in ORC and NRC see the discussion above(Vol. I Ch. 4). The letter in ORC is made in two parts, as illustrated in FIG. 11.4. The first part is known as the 'panse' (Vol. I Ch. 4), the second may be called the hasta68. These two parts are always made separately in the tablets, the 'panse' being made first and occasionally making a ligature with a preceding letter (e.g. ab, see No. 30.6 (295)); the hasta can ligature with a letter following, be (No. 22.16 (250)), bi (No. 37.24 (225)), br (No. 30.6 (295)) etc. In some tablets the two parts of the letter are almost equal in size, giving the letter a stunted, rather ugly appearance (e. g. Nos. 22 (250), 37 (225)); more often the hasta is tall and elongated, with the 'panse' reduced to a minimal curve, giving a much more elegant effect (e.g. Nos. 21 (248), 30 (295), 38 (346)). The letter can always be distinguished from d by the presence of a curve leading into the second stroke. There is no example in the tablets of b with 'panse à droite', the form in NRC, nor of any b which we should regard as closely resembling the 'baroque' form (see FIG. 11.5, Vol. I Ch. 4).

C. This is normally made, as shown in FIG. 11.6, in two strokes. Stroke 1 is always made first and can end with a slight curve to the right. It occasionally forms a ligature with the preceding letter (see examples under D and E). The second stroke, the cap, can ligature to the right, see e.g. ci in No. 37.22 (225), cr in No. 31.1 (297) and cu in No. 39.3 (299). Sometimes the cap rises upwards like later 'crested' c (e.g. No. 30), in which case there is no ligature; c can be made in a single stroke in a downward direction, but this does not often occur69. This form does not make ligatures either with preceding or following letters.

D. The basic form in ORC, as analysed by Mallon, occurs not infrequently in the tablets (see FIG. 11.7)70. The two parts are made separately, the loop first and the hasta second; the separation of the two parts can often be seen very clearly, especially in No. 30 (295) and in the rather idiosyncratic form in No. 4 (190). The loop can ligature with a preceding letter (we have noticed this only for A, see above), and the hasta with a following letter, as do (No. 30.5 (195)), dh (No. 37.11 (225)), ds (No. 4.9 (190) and 36 (223)) etc. Occasionally the two parts of the letter are written in one continuous movement, as in Nos. 1 (155), 7 (199), 22 (250) and especially No. 5 (191) where both forms occur (contrast lines 3 and 12). Alongside this 'uncial' form of d there occurs a second form, shown in FIG. 11.8, in which the hasta slopes to the right or is upright, as in modern lower-case d. Mallon considered the existence of these two forms of d as one important indication of the shift in the angle of writing in ORC71. It is therefore worthwhile noting that both forms occur in our tablets (for the form shown in FIG. 11.8 see Nos. 3 (160), 21 (248), 38 (346)) and indeed in No. 34 (218) we have both forms in the same text (contrast quid in lines 1 and 2 with ad in lines 3 and 4 etc.). Normally there is no way of telling whether d as made in FIG. 11.8 is a single stroke beginning at the top and ending with the loop or vice versa72; but in No. 53 (525), domine, the d ligatures with following o so that in this instance the letter must have been made in a downward direction ending with the loop.

E. Various forms of this letter are known in ORC73, all of which are found in the tablets. The form closest to capital, made in three strokes (see FIG. 11.9), is no doubt the oldest and occurs only rarely (e.g. No. 46 (342)). Most common is the form shown in FIG. 11.10. The left-hand stroke can be almost straight, or hooked at the foot, and it can be short and rounded (as in Nos. 42 (324) and 45 (341)) or slim and elongated (as in No. 30 (295)). In this form, made in two strokes74, it naturally ligatures freely with letters following, especially with r, er being the commonest of all ligatures in the tablets (e.g. No. 5.10 (191)). Among other examples note ea (No. 39.2 (299)), ec (No. 37.8 (225)), el (No. 38.8 (346)), em (No. 38.8 (346)), en (No. 30.6 (295)), es (No. 38.10 (346)), et (No. 4.31 (190)), eu (No. 37.6 (225)), ex (No. 37.21 (225)). Somewhat more surprisingly, this form can also ligature to the left, the join being made at the top of the left-hand stroke75: examples are cited under B, G, M, S, T and U. Note also an example such as meum in No. 37.6 (225) where e ligatures both at the left and the right. However, ei, which later became such a common ligature, is found only very occasionally in the tablets, e.g. in No. 32 (520) and in some instances of hordei in No. 4 (190). A more cursive form of e with the cross-stroke joined to the foot of the down-stroke is illustrated in FIG. 11.11. This is not uncommon in second-century papyri but is rare in the tablets (cf. No. 25 (247), 2nd hand). Finally, the form shown in FIG. 11.12, which became dominant in NRC but is found frequently in ORC in the third century and occasionally before this76, rather surprisingly seems to occur at least twice in the tablets, see No. 23.2 (263) and perhaps No. 77 (546).

F. Essentially there is only one form of this letter, as shown in FIG. 11.13. The size of Stroke 2 in proportion to the rest of the letter and the pronounced hook at the foot of Stroke 1 are the distinguishing marks between this letter and E. The form can vary in respect of its overall size (at times projecting well above and below the line, e.g. No. 21 (248)) and its position with relation to other letters (Stroke 1 can be very largely below the line of writing). Interesting variants on the basic form are to be seen in Nos. 22 (250) and 39 (299). The letter does not ligature with preceding letters but does occasionally ligature to the right by means of the cross-bar, e.g. fa (No. 31.7 (297)), fi (No. 37.22 (225)), fr (No. 5.9 (191)etc.) and fu (No. 37.3 (225)).

G. Again, there is only one basic form, for which see FIG. 11.14. This is normally made up of three separate strokes which are assembled in the way clearly illustrated by the g in axungiae (No. 4.35 (190)). Stroke 2 may be vertical (e.g. No. 22 (250)) or sloping, and varies in its length. Sometimes it seems to be made without removing the pen after completing Stroke 1, but there is no example of G made in a single stroke of the pen77. Stroke 3, the cap, is made last, as is shown by the way it can ligature with a following letter, see ge and gr in No. 30.1 (295) and 4. The letter does not ligature with preceding letters.

H. This occurs in a number of interesting forms, which is surprising as the letter is found only rarely in the tablets and has not been considered worthy of lengthy discussion by palaeographers writing about the letter-forms in ORC78. The normal form in ORC is shown in FIG. 11.15 and this is the one which occurs most often in the tablets. It can be preceded by a curve leading into the top of Stroke 1 (most notably in No. 4 (190)), and Stroke 2 can be made in two separate movements. Stroke 3 appears to have been added solely to facilitate ligature with a following letter but only rarely is such a ligature found in the tablets (see ha (bis) in No. 37.9 (225) and 11). Surprisingly, the letter once makes a ligature to the left79; see ad hanc in No. 37.11 (225). A further variant occurs in which the whole letter is made without removing the pen from the writing-surface (a form which also occurs in contemporary papyri); in this form Stroke 2 begins at the foot of Stroke 1, see coh in No. 30.3 (295); here, as one would expect, Stroke 3 has been dropped, but on one occasion (No. 37.25 (225)) it has been added to this form of h at its usual place, that is the shoulder of what would normally have been Stroke 2. It is unexpected to find two instances of h made in a form close to the capital, a form in which Stroke 1 does not project high above the cross-bar (as it does normally in ORC and in modern lower-case h), see mihi in No. 34.1 (218) and No. 39.3 (299).

I. This is an apparently simple and straightforward letter 80 and it is surprising to find it occurring in several interesting forms. The normal form in ORC is a short letter of the same height as m, n, u etc. (i.e. it lies within the 'corps de la ligne'); but ORC also permits an elongated form, known as 'long I', a form which persists into NRC. For 'long I' in the tablets see No. 40 (283) and especially Nos. 21 (248) and 22 (250), where both forms occur in the same texts. The more normal form is shown in FIG. 11.16. Here the noteworthy feature is the marked serif at the top, a serif which is present in at least half the instances of short i in the tablets. This is rather unexpected: i with a serif at this point does occur in papyri81, but it is far from common. This serif, especially if added just below the top of the hasta, can make the letter almost impossible to distinguish from p or, sometimes, from a or c. This serif facilitates ligature to the right, e.g. ia (No. 39.3 (299)), is (No. 30.5 (295)), iu (No. 39.4 (299)), but it is often used when no ligature follows (e.g. in Nos. 4 (190), 39 (299) and 43 (325)). Ligature to the left is very common. Note also the variant with a marked curve to the right at the foot found in Nos. 22 (250) and 42 (324). When i represents the numeral 'one' it often has a serif and makes a ligature, see especially the dates in Nos. 4 (190) and 7 (199); this is commonplace in contemporary papyri. For the strange form of i = 'one' in No. 30.4 (295) see the note ad loc.

K. This makes a ligature with a in kal, No. 30.8 (295).

L. The two basic forms in which the letter is found in the tablets are shown in FIGS. 11.17 and 11.1882. In the 'short' form (FIG. 11.17) it can have a pronounced serif at the top of the hasta (as shown), e.g. in Nos. 30 (295) and 34 (218). This form does not ligature with following letters and we have noted only one example of ligature to the left, with e in No. 38.8 (346) and 11. Note the interesting variant on this form in No. 45.5 (341). The form in No. 22 (250), with a hook leading in to the hasta, which is somewhat elongated, is probably best regarded as a variant of this form, although it is quite as tall as some of the instances of the form shown in FIG. 11.18. This 'long' form can in fact vary considerably in length. It is the commoner of the two forms and can be a slim, elegant letter, made almost exactly like b without the 'panse', cf. Nos. 21 (248) and 30 (295). Both forms are found in No. 31 (297). We have not found any example in which this 'long' form makes a ligature to the left83; for ligatures with following letters cf. la (Nos. 37.14 (225), 39.1 (299) etc.), li (No. 37.19 (225)) and lu (No. 37.14 (225)). In papyri, by the second century, it is very common to find l made with a long diagonal for the foot, sloping down well below the line. Surprisingly, there is no good example of this in the tablets, though in ceriali (No. 24.1 (270)) and the first l of luguualio (No. 22.9(TVII)250) we have something approaching this form.

M. This letter has been analysed several times and Mallon distinguished four variations, all of one basic form, in which the letter is made in one, two, three or four movements of the pen84. This analysis seems to us to be correct and we suspect that all four variations are present in our tablets, although it is very difficult to be sure. Indeed, all four variations may be found in Nos. 4 (190) and 37 (225). The form made in four separate strokes, shown in FIG. 11.19, can clearly be seen in Nos. 34 (218) and 48 (522). The form shown in FIG. 11.20, in which the middle two strokes of the previous form are linked together in one movement, is to be seen in all probability in No. 46 (342) and there is an interesting variant of this in No. 42 (324), where Stroke 3 of the form in FIG. 11.19 has almost totally disappeared. The form in FIG. 11.19 can be modified slightly by making everything after Stroke 1 in a single movement of the pen. In this form Stroke 1 is exactly like Stroke 1 of a in FIG. 11.1 and, like this a, ligatures freely with preceding letters (see examples under A, E and u). When this form ligatures to the right, which seems to be uncommon, an extra linking stroke is added, cf. ma and me in No. 37.6 (225) and 14-5. When m is written entirely in one movement (still in the basic form shown in FIGS. 11.19 and 11.20), as probably occurs frequently in the tablets, it can ligature with a following letter. There are particularly instructive examples of this happening with i and u in the second hand in No. 21 (248) and with i in amicorum in No. 37.17 (225); here the second half of the letter is distorted, with the final stroke being made in an upward direction to facilitate the ligature85. Unless this can be regarded as a precursor of the form usual in NRC, which does not seem likely, there is no example of m in a form akin to that of NRC. Finally, we should note two examples in which the final stroke, at the end of a line, is prolonged with a curve: see No. 43.2 (325) and especially No. 30.4 (295).

N. This is often found in the basic capital form, sometimes heavily serifed (e.g. No. 34 (218)), and in this form it is never found in ligature. This basic form is made in three movements, but variants are found in which the second and third strokes can be combined into a single stroke (as in FIG. 11.21), or in which the whole letter is made in a single movement, beginning at bottom left and ending at top right; in these variants the direction of some or all of the strokes differs from that in the basic capital form86. Note also the variations on this form in No. 22 (250), where the initial stroke sometimes extends well below the line, and in No. 32 (520) where the last stroke has a flourish ending in a curve. In this form it seldom makes a ligature; cf. however en (above under E), ni and nu in No. 31.3 (297) and 1, and (probably) ns in No. 43.6 (325) and nt in No. 23.5 (263) and No. 38.10 (346)87. A notable variation on this basic form is shown in FIG. 11.22, in which the cross-stroke begins a long way to the left of the first hasta and meets the second hasta at or near the top (e.g. Nos. 2 (151), 4 (190) and 48 (522)). In No. 42 (324) the cross-stroke and second hasta are joined in a way which seems to show the letter on the way to the form illustrated in FIG. 11.23, a form found in Nos. 5 (191), 25.2 (247) and especially No. 37 (225) (and elsewhere). In No. 37 (225), in fact, n occurs in several variations, the most interesting being those where the final stroke ends in an upturned loop and that in for instance beneficio (line 22), where the letter seems to be made in two strokes only and to resemble the type of n which was to appear in NRC. These variations occasionally make a ligature either with preceding letters (an, No. 37.11 (225); en, No. 5.3 (191), No. 21 (248) 2nd hand, No. 23.5 (263)) or following letters (no, No. 37.2-3 (225)).

O. Although the letter can be made as a simple circle, it is nearly always made in two sections as shown in FIG. 11.2488. Clear examples of the two halves are to be seen in No. 4.36 (190), etc. and in the ligatures to the left (examples under D, E, U) or the right: on (No. 37.16 (225)), or (Nos. 4.16 (190), 30.4 (295), 37.17 (225), 38.4 (346)), os (Nos. 6.1 (205), 37.18 (225)). When it ligatures to the right, which happens only rarely in the tablets89, the form of the letter is usually noteworthy.

P. FIG. 11.25 shows the usual form of this letter in ORC and in the tablets. Variants are found at the foot of the hasta and in the angle at which the cross-stroke joins the hasta. The foot may be left plain or may curve to the right; it has a pronounced additional stroke in No. 22 (250). The cross-stroke may slope downwards as shown (which seems to be the most 'correct' form) or it may be horizontal, or even slope upwards; in these cases the letter becomes almost impossible to distinguish from c or, at times, from i (with serif) or t. Ligatures follow the pattern indicated above under C, but are rare with preceding letters and not common with those following except for pr (Nos. 5.4 (191), 37.8 (225) etc.); cf. also pe and pu (No. 37.21 (225) and 8), and ps (No. 58 (529)). Note also the elongated version of this form in No. 5.6 (191). Alongside this form there appears a form very much closer to capital and resembling modern P, that is with a loop or bow at the top (usually closed): this form is much less common (see Nos. 2 (151), 42 (324), 45). In contemporary papyri such a form is found not uncommonly. Casamassima-Staraz90 describe it as the 'posed' form and it is this form which they believe was important for the development of NRC.

Q. There are two basic forms of this letter in the tablets, as illustrated in FIGS. 11.26 and 11.27. It is not clear whether either form is ever made in a single movement or whether the loop or circle is always made separately from the 'tail'. No examples of ligature have been noted. It has been argued that the form in FIG. 11.27 is a later type than that in FIG. 11.26 and did not come into use until after the change in ORC which took place in the early second century91. It is therefore important to note that both forms are in regular use in our tablets.

R. The basic form of this letter is shown in FIG. 11.28. Stroke 1 can vary in length, direction (vertical or sloping to the left) and in the presence or absence of a hook or curve at the foot; note especially the very long first r of tungrorum (No. 30.4 (295)) and the marked sweep to the left at the foot of r in ostria (No. 39.3 (299)). Stroke 2 can show a number of variants, some of them so considerable that they might suggest a different form of the letter rather than a variant on a single basic form92. Commonest is the 'wavy' form, as illustrated, but this 'wave' can be almost or entirely lost, and we may have only the second half of it with no part of the second stroke made to the left of the top of Stroke 1 (e.g. Nos. 21 (248) and 24 (270)). When the 'wave' is lost it becomes difficult or impossible to distinguish the letter from a. Interesting forms are found in No. 42 (324), which clearly shows the derivation of the letter from the capital, and in No. 4 (190) (see the introductory comments and cf. also No. 7.3 (199)). The ductus of the strokes is essentially the same as for a. Ligatures with preceding letters are very common, especially the ligatures er (see above under E). For ligatures with following letters cf. ra (Nos. 30.6 (295), 34.4 (218)), re (No. 37.15 (225)) and ri (Nos. 31.1 (297), 37.14 (225)).

S. Again there is room for doubt whether all examples of s are really variants of a single basic form (illustrated in FIG. 11.29). On the whole we think this is true of the examples found in the tablets93. The principal variation is in the size of the letter: it can rise well above and descend well below the line (e.g. Nos. 21 (248), 30 (295)); or Stroke 1 can keep within the size of other letters while Stroke 2 rises well above them (the most usual variant); or the whole letter can be made short and within the 'corps de la ligne' (both of the last two variations occur within the space of a few letters in No. 10 (224)). Stroke 1 can be just a straight descender or can curve markedly to the left at the foot; it seems always to be made from the top downwards and, as it frequently ligatures with letters preceding (see examples under A, D, E, I, O, P and especially U), it must have been made first. Stroke 2 is normally and perhaps always made from the left to the right. In the shortest variation of the letter this stroke can be horizontal or raised only very slightly; in this variation it can ligature to the right, e.g. sa (No. 30.5 (295)), se (Nos. 30.6 (295), 31.5 (297), 41.1 (323)). Note in particular the ligature si, where i is in the 'long' form, and as in occassionem (No. 37.16 (225)); see further ss in esse and missit (No. 39.2-3 (299)) and st (No. 23.3 (263) and 5)94.

T. The normal form in ORC and in the tablets is shown in FIG. 11.30. Cencetti, who argues that the letter in ORC is palaeographically of more interest than might be thought95, has detected examples in the papyri of t made in the form shown in FIG. 11.31, that is with the left-hand half of the cross-bar joined to the hasta and the right-hand half added afterwards as a separate stroke96. It is doubtful whether the tablets contain any example of this form; possibly it is to be seen in No. 1.8 (155) and again in No. 31.2 (297), where t ligatures with following e97. In its normal form the hasta is made first and the cross-bar second, as is shown by the fact that ligatures with following letters are always made with the cross-bar: e.g. ta (No. 39.4 (299)), te (Nos. 37.2 (225), 39.2 (299), 43.3 (325)), ti (Nos. 31.9 (297), 37.10 (225)) and tu (No. 4.22 (190)); note especially the ligature in tamen in No. 41.2 (323) where the cross-bar slopes downwards to facilitate the ligature and the letter looks like p in FIG. 11.25. Alongside these expected ligatures there are a few instances where t apparently ligatures with a preceding letter by means of the cross-bar: especially clear is at in No. 41.1 (323), where a link stroke is added; cf. also nt (No. 38.10 (346)) and st (Nos. 23.3 (263) and 5, 39.1 (299)). If these are true ligatures, the cross-bar must have been made first in these instances.

U. The letter can occur in several forms, three of which are illustrated in FIGS. 11.32-498. Commonest is the form shown in FIG. 11.32, either written on the line or slightly above it. It is regularly above the line when making a ligature to the right, such ligatures being common, e.g. with a (No. 38.11 (346)), e (No. 37.6 (225)), i, m (No. 37.17 (225)) and s. Often the right-hand part of u coalesces with the following letter, especially in the ligature us (no doubt deriving from the conjoint us of stone inscriptions): for u in us reduced to no more than a slight hook see, e.g. Nos. 21.9 (248), 43.1 (325) and cf. No. 38.10-11 (346)99. Instances of ui ligatured in such a way as to resemble q with a straight descender are not common: cf. cuius (No. 37.11 (225)) and perhaps qui (No. 40.5 (283)). Ligatures with preceding letters are unusual, whether u is made in this or one of its other forms. Tablet No. 4 (190) has u in at least three forms: the one just discussed, a variant where u is larger and made in the shape of a deep bowl (cf. line 40, uini) and a more angular form illustrated in FIG. 11.33 (cf. line 25, uini and, regularly, the numeral 5; see also No. 7.2 (199)). A different form is shown in FIG 11.34, where u has a hasta at the right and resembles modern lower-case u (e.g. Nos. 22.10 (250), 37.19 (225) (plurimos), No. 46 (342) (with marked serifs) and No. 47 (152)); there is a slight variant in No. 37.24 (225) (uindolanda) where the hasta comes below the level of the rest of the letter. Sometimes this hasta curves to the right at the foot (e.g. No. 37.9 (225)) and very occasionally in this form it can ligature to the right as in uo (No. 37.18 (225)). Finally, note the remarkable ur in muriae (No. 4.33 (190)), perhaps best described as a conjoint letter (see note ad loc.).

X. Probable ligatures are to be seen in ex in No. 37.21 (225) and 24, and perhaps in xs in No. 37.21 (225). The letter calls for no special comment.

Y and Z do not occur.