CH. VII. added to a verbal base in e, as ferue-sco, &c.: yet even here sometimes both forms are found; we have conticiscam, luciscit, &c.1 An important example of the attraction of s alone is furnished by the termination of comparatives, -ius for older -ios. Here the i was sometimes absorbed by the u, as in minus: but more commonly the u itself sank to i, which then coalesced with the preceding i, and should therefore have produced a long final syllable in magis, satis, &c.; but the weak pronunciation of the last syllable in Latin seems in every case to have let the long vowel pass away. Lastly, i occurs in connection with t and d. The instances where this assimilating power is best seen are the participles or participial formations from the second conjugation, as meritus from mere(but merētod occurs in the well-known epitaph of the son of Barbatus), tacitus from tace, and others too many to quote. The same change is seen sometimes, though rarely, in the first conjugation, as domitum, cubitum, crepitum, uetitum, &c. In these cases accent no doubt had much to do with the weakening: the unaccented middle syllable could not maintain its length, and the shortened vowel easily sank to i. The fact that i is always found before the suffixes -tion, -tia, -tāt, -tudin, -do, &c., should perhaps not be pressed as an instance of assimilation, for we have already seen that in all such formations the final vowel of the base has a natural tendency to sink to i as the easiest vowel, e.g. in belli-cus, rubi-cundus, &c. In spite of the tendency to e before two consonants we find i when both are dentals: e.g. intus (èvтós), indu, for the endo of the XII. Tables, uindico (mentioned above) and uindex, uindemia, &c.: also before gn in ignis, tignum, signum, pignus, which is some slight ground for believing that the g here was the guttural nasal : however the same change is seen where the g follows the n in lingua, tinguo, Inguis, &c. and seems due here to the n alone. 3 1 Plaut. Mil. 410; Terence, Heaut. 410. 2 Corssen, II. 299. 3 At page 264. In conclusion then, the three weak vowels have their own peculiar affinities; u for l and labials, and e for r and closed syllables; i for dentals: these affinities being the result either of the difficulty of other combinations or of the nearness of the point of contact of the two sounds. Next we have to consider the cases where one vowel has assimilated another. This phenomenon is of comparatively rare occurrence. We have often had occasion to remark the weakness of the Latin vowel-system: consequently we shall not expect the vowels to exercise so strong an influence over another vowel as the consonants did. The following examples are derived, as before, almost entirely from Corssen. (i) When two vowels come into actual contact, they have a tendency to approximate to each other. Thus when y was resolved into i in (e)syam, the subjunctive of √es, the difference in point of distance between the two vowels i and a made the form siam unpleasant: hence, through the influence of the i, the a drew one step nearer to it, and became e-siem, a form which constantly occurs in Plautus'. That a really occurred in this form in Latin, as well as in Graeco-Italian or Indo-European, is shewn by those cases where the y was entirely dropped, not resolved into i, when the a remained intact; as in reg(y)am. So also we have eam from i, queam from qui -where the radical vowel is changed. We have already seen that the Italians kept e in many words where it sank to i in the Latin: with the same preference for that sound they assimilated i to e in the common termination -io: thus Corssen quotes fileai from Praeneste, at an earlier date than 218 B. C.; not forty years later than the filios of the younger Scipio's tomb. Similarly, the difficulty of the combination ia produced the numerous class of secondary nouns in -ies, e. g. durities by the side of duritia. Another effect of this assimilating influence of one vowel on another is to check in some cases the same power CH. VII. 2. Vowel assimila tion caused by vowels. CH. VII. when exercised by a consonant. We have seen already that o followed by almost always sinks to u. But this change does not take place when i or e precede o: apparently the labial action for u was felt to be inconsistent with these two vowels; and they therefore by their assimilating power retain the original o in uiola, filiolus, gladiolus, &c.; in aureolus, luteolus, &c.1 (ii) When two vowels are separated from each other by a consonant, they tend to become identical. Thus e assimilates a preceding vowel in bene, originally bono, which by regular weakening became bone; then the feeling of the coming e in the last syllable modified the o in the first. Similarly illec-ebrae owes the e of its second syllable √lic to that of the third. O has changed u and e in a previous syllable, in soboles (sub) and socordia (se). So also u has operated in the suffix of tug-urium on the vowel of √teg, and perhaps caused partial assimilation in so-luo, so-lutus for se-luo, like se-cors; we find lucuna for lacuna, and rutundus in Lucretius. I think it possible that the same influence may have produced diurnus (dies), and arbustum (arbos). But more numerous are the cases where i has affected a preceding vowel, Thus ne-hilum becomes nihil: the old i in mihi is preserved by the final i, though in mei, meus, &c. it has become e: √sul in consulo passes into sil in consilium, facul becomes facilis; semol (simul) becomes similis: and cal, which is found in KaλÚTтw and calim (the old form of clam according to Festus), becomes occulo, but super-cil-ium: Caecus becomes Caecilius; and contrast Proculus with Procilius, Lucullus with Lucilius. I do not think that inquilinus by incola, inspicio by spec, can be fairly quoted as instances3: because the vowel would have in each case sunk to i by itself. A forward action is clearly to be seen, as I think, in difficilis and displicet*: i is not found in perfacilis and perplacet, where no i precedes. It appears from these examples that by far the greatest part in this kind of assimilation is played by the vowel i-the weakest of all: a fact which is certainly surprising. Corssen1 gives the analogy of ä, ö, ü in German, which are commonly produced by an i in the following syllable: e.g. mann, männlich: and he concludes that i, thin though it be, requires for its pronunciation a considerable tension of the organs of speech, differing herein much from e. This explanation is most unsatisfactory. It is this effort required in pronunciation, and nothing else, which is the mark of a strong vowel: and yet nothing can be plainer than the fact that i is weaker than a, o, or u. And certainly no such tension is absolutely required to sound the i, though greater power may accidentally be applied to it, as it may also to e. The truth is that the real cause of the change is not the influence of the i: the real cause is the natural tendency of every vowel to grow weaker in Latin: the i only lends a helping hand, determining how far the change should operate-in this case to the utmost possible limit, sometimes giving an additional impulse to the vowel affected, which might otherwise have resisted the primary tendency, as difficilis, mentioned above. In a word, it is only a modifying, at most an auxiliary cause of the change: and this is in accordance with the view of Assimilation which I have given. Corssen' gives some interesting examples of a produced by assimilation in the late popular-Latin: e.g. ansar for anser, parantalia, &c.: and he points out how a in this way appears sometimes in the Romance languages, e.g. marchand from late Latin marcator, sauvage from salvaticus (silua). It seems to me unquestionable, that this a, so produced, was not the full sound (ah) in Latin: though it may have become so in the descendants of the Latin, all of which, as has been pointed out, were subjected to foreign influences. It may have been (ă), but more probably the neutral vowel3. It is observable, that in almost 3 See page 84. 1 II. 380. 2 II. 373. CH. VII. Apparent influence of the vowel i. CH. VII. all cases this a precedes an r, or l, that is, just the two sounds before which the neutral vowel is most common in England, e.g. altar, fatal. Less frequent in its operation acts principally as a bar to further change. III. DISSIMILATION. This principle has of course a less wide field than that which we have just considered. The same sound is less likely to occur twice in inconvenient proximity, than different sounds. Like Assimilation, it is sometimes an auxiliary cause of new change, sometimes it prevents the regular process of change. Its operation is restricted to some of the places in which either by regular substitution, or by the loss of a letter, or by the resolution of a semivowel into a vowel, or by the addition of suffixes to roots or bases, or by two of these causes combined, the same vowel-sound occurred twice. It acts, I say, only in some of these places, because the most obvious method was to let the two vowels so meeting coalesce into one long vowel and this often took place. For example, when sequ-ontur was tending to become sequ-untur by the regular substitution of u for o, since the double u would have been difficult to pronounce, the two often coalesced, and ( being rarely written after the loss of its peculiar attendant u) the result was sec-untur, when the tendency to weaken o to u in these forms had become too strong and too universal to be resisted. But the natural dislike to such a transformation is seen in the fact that the old spelling sequontur was still retained even in the Augustan age, side by side with the new. Similarly we find in indifferent use equos and ecus, aequom and aecum, quom and cum, &c. In all these cases this retention of the o, this bar to the regular change, is due to the principle of Dissimilation. In some instances no doubt this principle was aided by another cause. If the weakening of o to u had taken place, and the two vowels had then coalesced, |