:

is more intelligible if the sound was w: the name by ana-
logy should have been ev (for, in momentary sounds, the
vowel follows, in continuous sounds it precedes the conso-
nant), and, doubtless, would have been so if the sound had
been v: but it would have been inaudible, or nearly so, if
the consonant was w; and, therefore, the vowel was in this
case post-fixed. The above arguments are all Mr Roby's:
to him also is due a most thorough discussion of the trans-
literation of v into Greek: the fact, that B is often found
there for it, is the strongest argument for the v-sound.
Two questions here arise: what was the sound of B? and
how far was used? In reply to the first, ẞ is certainly
v in modern Greek; but this v seems to be labial, not la-
bio-dental and I have already pointed out the affinity of
the labial v and B. Consequently, if ẞ had represented v
regularly in Greek transliteration of Latin words, this
would be an argument only for the labial sound of v, that
is, for something nearer our w than our v.
But the repre-
sentation of v by B, as Mr Roby has clearly shewn, occurs
but rarely in the older writers, and with increasing fre-
quency the later we pass along the scale. Thus, in Poly-
bius (second cent. B.C.) ou is regular: in Dionysius Halicar-
nassensis ou is much more common, but ẞ is also found:
but in Plutarch (first cent. after Christ) the number of
times in which B occurs is 180, while ou is still found 323
times. It is clear, therefore, that this transliteration of v
by B cannot be separated from the tendency to confuse to-
gether the v and b in Italian itself, which shews itself in
the second century after Christ, and afterwards increases:
but which probably occurred, at least dialectically, much
earlier, if we may judge from double forms, such as Labici
and Lauici, Fabius and Fouius, &c.: it is by far the most
probable that the v, in all such cases, was the labial v, which
passed into b irregularly, but never permanently: thus,
uiuere was bibere in inscriptions of the fifth century after
Christ, but the v is found again in modern Italian. The
occurrence of a labial v in Latin would also account for

CH. VIII.

CH. VIII.

rare changes, like ferueo and ferbui, boues and bubile, &c. Therefore, so far as the argument from transliteration goes, we seem to have evidence for a w-sound, or a labial v: and, combining this argument with those already adduced, I think it most probable that v was generally w, but sometimes (dialectically) a labial v.

The Latin aspirate f.

4. Changes of the Aspirates in Latin.

Lastly, I shall take under the head of Substitution the numerous changes of the aspirates in Latin. Some indeed of them seem to be due to Loss: others, if Corssen's explanation of them be true, should rather come under the head of indistinct articulation. But since neither of these causes can be certainly made out, and since, if divided, the history of the aspirates would be less intelligible, I have thought it better to put the whole of the changes together under the simplest head at all events one sound has been substituted for another.

The most remarkable point in the history of the aspirates in Latin is that each of them can be represented by one symbol, the peculiar Italian f. That this ƒ is no aspirate is obvious, if only from the fact that it has not the power of the Latin momentary sounds to assimilate a nasal which precedes it: we have im-petus for example, but only inficio1: this shews that the ƒ is quite different from the Greek 4, which has the assimilating power, as in èupaivw3. For the difference of sound between these two, we have Priscian's well-known dictum: that ph is produced "fixis labris," but f not. This need not necessarily mean that ph was a momentary sound (though it probably was) and f a protracted one: it may only mean that ph was a labial

1 Corssen however (1. 138) quotes some examples from the Corpus, as com-fluont, im fronte. But these are certainly exceptional.

2 See Curtius in the Zeitschrift, 11. 333.

and ƒ a labio-dental: which last statement is probable on other grounds.

It regularly occurs as the representative of initial BH. This we should expect from its partially labial character. Thus we have fari from BHA, whence Greek pá-vai; fui from BHU; fugio from BHUG, &c. But it is hardly less frequently found as the representative of initial DH. That aspirate has left no Latin exponent of its own kind, at once dental and aspirate, or even a dental spirant: ƒ has taken the place. Thus fumus is the Latin derivative of DHU, the same in form as Ovuós and Sanskrit dhuma; fores represents dvára (Sk.) and Oúpa: firmus is from DHAR "to hold firmly;" a root which gives an extraordinary number of derivatives in Latin', including formido "stiffening fear," forma, forum, and many names of "strongholds," as Formiae, Ferentinum, Forentum and Ferentia: many more examples are given by Corssen. Both the labial and dental aspirate are regularly represented at the beginning of a word by f.

CH. VIII.

F represents original BH

and DH at the beginning of a

word:

even GH.

But there are even cases where initial ƒ represents GH. sometimes Such are fel, which seems undoubtedly to be the same as xoλn, our "gall," that is from original ghal; the verbs quoted by Priscian and Festus, futire and futare, with futilis are from fu, the same as √xʊ in xeƑw; formus and feruor stand by Sanskrit gharma, our "warm," while the Greek shews a change from the guttural to the dental in lepμós: fames, and ad-fatim fatigo are from the same root as xá-Tis; friare and friuolus are akin to xpí-ew. Commonly however there is another form beginning with h, existing side by side with that in f, and used by educated men"; we have faedus, but classical haedus, our "goat," where the Teutonic has kept the g of the original GH: fordeum and hordeum, German gerst; fariolus and hariolus, Greek Xop-dń; folus and holus, Greek xλoń; fostis and hostis, German gast, our "guest;" &c. This f for gh is only initial.

P. E.

1

I. 148.

2 Krit. Beitr. 212, &c.

22

CH. VIII.

In Latin

the h was commonly dropped in the middle of a word.

If we continue our search, we shall find that this ƒ does not occur much in the middle of Latin words. We have scrofa, the pig, conceived as the "grubber," by the side of scrob-s1, and probably scrib-ere. But, as a rule, we shall find that BH has almost universally under these circumstances passed into b: e.g. ambo, tibi, lubet, nubes, &c., with hosts of others. But it is an instructive fact that by the side of the Latin b there is found ƒ in the other Italian dialects. Thus Safinus is the Oscan for Sabinus; the proper names Alfius and Alfenus should be compared with the Latin Albius and Albinius: Orfius with Latin orbus, &c. The same mutatis mutandis applies to DH: this is d in Latin medius (madh-ya), in aedes, in dedo, condo, &c., from DHA "to place," &c.: but the Oscan for "middle" is mefia; and the Oscan Rufium, with the proper names Rufus, Rufinus, &c., seems to shew that rufus “red” was borrowed by the Romans, their own word being ruber. The root from which the two forms came is certainly RUDH, the Sanskrit rudhira, Greek è-pve-pós, our "red:" ruber therefore shows us that in Latin b can represent medial DH; as we see also from uber (ovlap, "udder")—but Ufens, Aufidus, in different parts of Italy; from uerbum, a “word,” and barba, a "beard:" in these last two words the traces of original DH are preserved by the Teutonic languages with great fidelity, and by them only.

It is clear then that both DH and BH were regularly represented in Italy by f: though the Latin alone preferred the more distinct d and b within a word. At an early period the DH must have passed into bh in Italy: so that from original rudhira came the old Italian rubhro, which then split into Italian rufru and Latin rubro just on the same analogy (as Curtius points out') as old Italian tibhi (where BH is original, compare Sanskrit tubhyam) split into

Umbrian tefe, Latin tibi. This weakening of dh to bh is CH. VIII. neither impossible nor unnatural: we have already seen how inexactly d was sounded in Latin, so that it could pass into both land r. But I think we may believe that the breath at the end of each aspirate was somewhat strongly sounded in Latin, so that the distinction between the b and the d was not appreciable, and therefore they sank to the same spirant f. This view appears to me to be supported by the fact that ƒ from bh sometimes passes into h: as in harena for the old Italian and Sabine fasena: haba exists by the side of faba: herba is most likely from /bhar, compare popß; and mihi undoubtedly stands for mibhi, the loss of the b being possibly due, as Curtius suggests, to the dissimilating influence of the labial m. Now there are tolerable indications that h was a strong sound in the old Latin: although in the Augustan age no doubt it had grown weak, and was constantly dropped, as in (h)anser, (h)olus, &c. But the strength of the breath in former times, when the changes between different classes took place, would be a good reason for the change between strong hand f with a strong breathing. And the same conclusion may be drawn from the occasional substitution of ƒ for GH mentioned above. I pass now to the more regular changes of GH, to complete the history of the aspirates.

GH is generally represented by g when not initial. Changes of Thus ang-or is from AGH, whence axos, &c.; lig-urio is from GH. LIGH (λeixw); so also anguis, unguis, ningit, &c. have lost the h3; when it stands at the beginning of a word as in gramen, granum, grando, &c., it seems to be generally followed by r, which absorbed the breath but left the g1. Initial GH is regularly represented by h; as hiemps (GHI, whence Xi-úv, &c.), heri (Sanskrit hyas for ghyas, Greek xoes where the is peculiar"), hostis (from GHAS, whence our "guest:"

1 Corssen, I. 102. 3 Comp. 245.

5 See Gr. Et. 454.

2 Id. 1. 106, &e.

4 See Grassmann, Zeitsch. XII. 89, &c.