This paper is intended as a short linguistic contribution to the study of an important feature of Indo-European religious history: the sacrifice of the horse. There exists a copious and valuable bibliography on the subject, and I shall refrain from needless repetition of the materials and conclusions presented there. Only features of immediate importance for my comparison will be explicitly mentioned.

In 1858, a farmer plowing a field in Suszyczno, near Kovel in the old administrative unit Volhynia (now in the northwestern Ukraine), turned up an iron spearhead with an inscription and abundant ornamentation. After its publication some seventeen years later, the inscription became, and has remained, an object of keen interest to scholars, who at first had been inclined to doubt its authenticity because of forgeries that had recently occurred. R. Henning (who saw the spearhead itself before it was lost), L. F. A. Wimmer, and others were convinced that it bore inscribed runes and that the inscription was East Germanic. This is still the general opinion; because of the final -s of the inscribed word, it is considered to be Gothic. It is believed to date from about the 3rd century A.D.

In a recent article, Lg. 29.308 (1953), Reichardt discusses the chronology of syncopation of the connecting stem-vowel in harigasti, a name contained in the inscription on helmet B of Negau. Reichardt's argument raises an interesting combined point of method and of fact. From attested old (and dialectally unspecified) Germanic names, of which Reichardt adduces a few examples, a seeming anachronism presents itself in the occurrence of compounds with -io-as late as the 4th or 5th century, while in harigasti and in Harimellae we seem to find *-ja- reduced to -i- at a date considerably earlier than the 4th century a.d. It is, of course, possible to allege that in each case we have forms from a different dialect, and that the change occurred separately at different times in different places; but it is unsatisfying to argue from our ignorance of prehistoric Germanic dialect distributions, and Reichardt wisely refrains from introducing this answer of despair.

In contrast to other IE languages, Armenian shows a great variety of developments from IE voiceless stops; only IE *k̂ appears to yield Arm. s consistently, and will therefore be disregarded in the present discussion.

[An Indo-European cluster composed of an originally initial voiced laryngeal and an immediately following */y/ coalesced in Armenian to yield the voiced palatal affricate /ǰ/. This development resembles a similar change in Greek.]

The following investigation presents a constructional procedure segmenting an utterance in a way which correlates well with word and morpheme boundaries. The procedure requires a large set of utterances, elicited in a certain manner from an informant (or found in a very large corpus); and it requires that all the utterances be written in the same phonemic representation, determined without reference to morphemes. It then investigates a particular distributional relation among the phonemes in the utterances thus collected; and on the basis of this relation among the phonemes, it indicates particular points of segmentation within one utterance at a time. For example, in the utterance /hiyzkwikər/ He's quicker it will indicate segmentation at the points marked by dots: /hiy.z.kwik.ər/; and it will do so purely by comparing this phonemic sequence with the phonemic sequences of other utterances.

In grammars of Germanic the reflexes of the PIE resonants, /y w r l m n/, are presented either inadequately or unsystematically. Prokosch devotes but a few pages to them, in contrast with his thirty-odd pages on the reflexes of the obstruents. Streitberg deals with them somewhat more fully, at various places in his treatment of Germanic phonology, but his chief interest is in the assimilatory and dissimilatory changes in which they are involved. The relative neglect of the resonants is understandable, for until Edgerton described in detail the allophonic distribution of the PIE resonants, it seemed completely adequate to cite an example or two of the correspondences between PIE [y w r 1 m n] and PGmc. [y w r 1 m n], or PIE [i u] and PGmc. [i u], and then to state with examples that a [u] developed in the neighborhood of PIE syllabic . If however we accept Sievers' law in Edgerton's formulation, we cannot pass over the Gmc. developments of the PIE resonants so lightly. Since we now expect grammars to present full and systematic statements of the general lines of development of a language, future treatments of PGmc. will have to describe in some detail the developments of the resonants into Germanic. In this paper the general outlines will be sketched, especially those environments which have been inadequately treated in the past.

The dialect area with which this study deals is the Roman imperial administrative district of Narbonensis, the southeastern part of Roman Gaul—in essence, the modern Provence. A more exact delimitation of this region, together with a map, may be found in DAG pp. 59-62 and 14. All the extant evidence of the Gaulish language has been collected and classified by What-mough in DAG; with one or two exceptions from CIL 12, the material for this study has been drawn in its entirety from the section on Narbonensis, DAG 20-83.

In historical English grammar it is regularly assumed that Gmc. initial [f s þ] became voiced before voiced sounds in southern England at some time in the Old or early Middle English period. For the present purpose, Wright's description of the change is sufficiently representative: ‘The initial voiceless spirants f, s, þ became the voiced spirants v, z, ð in late OE. or early ME. in Kentish and the southern, especially the south-western dialects, as vader, vat, vlesch, vrend; zaule zǫule, zinne zenne zünne “sin”, ðat ðet, ðing. The modern dialects show that this voicing of the initial voiceless spirants must have taken place at an early period, because it is almost exclusively confined to native words, hence the change must have taken place before the great influx of Anglo-Norman words into these dialects. The use of the initial voiced for the voiceless spirants is now obsolete in Ken., Sur., Sus., and obsolescent in s. Pem., Hamp., and the I.W., but it is still in general use in east Hrf., parts of Glo., west Brks., Wil., Som., and Dev.‘

The purpose of this paper is to advance a theory as to the circumstances, linguistic and cultural, in which the uvular [r] has been substituted for the denti-alveolar trill in modern French.

In 1911 the Abbe J. Rousselot wrote: