> ué,2 and the Celtic theory of Goidanich that the development was ŏ > uó > uö > ue.3 With respect to the history of accented ě, which is in general the same as that of Ŏ, one finds also new data: the frequency of ia in Leon, the antiquity of ie in Castile and Aragon (ninth and tenth centuries), the presence of both ie and the archaism e in Mozarabic, etc. Of special interest is the evidence presented to show that the development ě > ie > i in the suffix -ěllu took place in Castilian in the tenth century despite the fact that in the formal literary documents it does not appear generalized until the fourteenth century. Menéndez Pidal believes that the presence of ié in the written language until the fourteenth century in Castilian may be due to Leonese influence. The statements usually made by philologists (Bourciez, Zauner, Hanssen, Baist, etc.) that this change took place in Castilian at the end of the thirteenth and at the beginning of the fourteenth centuries are therefore wrong. Of the greatest interest and importance for Romance philology is the discovery and study of a series of phonetic forms that reveal the presence of a third written language in Leon during the tenth century used side by side with scholastic Latin and popular Leonese, §§32-34. According to the conclusions of Menéndez Pidal these forms are for the most part archaisms that continue Mozarabic speech such as was spoken and written by the Visigothic Spaniards in the beginning of the eighth century. Here belong: (1) words that retain the posttonic vowel with regular development of the accented vowel or with regular voicing of the intervocalic voiceless explosive, siéculos, cómide, pacifigas; (2) forms such as the above but with popular development also in the posttonic vowel, 'vocal con timbre vulgar', quéncoba < concoba, púdeda < putida, or with a changed vowel and different (though phonetically equivalent) consonant, diabulo, ribolus, tabola; (3) forms that retain the pretonic vowel and voice the medial voiceless explosive, pretonic unchanged, comidesa, eredidade, vowel with 'timbre vulgar', leterado, pobolatos. The evidence submitted in §35 for the primitive difference in development of final ō and u tends to show that the generally accepted view that these were blended into a closed o in Vulgar Latin does not hold. A. Gl. Itai. 4. 402. The explanation given by Menéndez Pidal for the presence of úo, úe in Leonese and in Sicilian in the so-called cases of 'realce acentual y énfasis' does not explain the final definite u and i < Latin accented ǎ and è in Sindanés (south of Miranda do Douro). See Leite de Vasconcellos, Estudos de philologia mirandesa, Lisboa, 1900, §§50, 58. L'origine e le forme della dittongisazione romanza, Halle, 1907, pp. 36ff., and Meyer-Lübke, Grammaire I §211. In Spain final ō appears as o and final u as u in the tenth and eleventh centuries frequently enough to make one suspect that the development of ŭ to o in this case was rather late in Castilian and in Mozarabic. The majority of the cases of anaptyxis given in §40, vowel introduced between cons. + l or r, and between l or r+cons. are really covered by the rule that holds for the presence of the same phenomena in Oscan' and makes one suspect that in Castilian origins the influence of the old Italic dialects may be even greater than Menéndez Pidal has supposed in §§52-54. The study of the history of Latin initial ƒ in Spanish, §41, is of transcendental importance. Menéndez Pidal has found abundant evidence to show that the territory where f-> h- in the eighth and ninth centuries (the earliest Spanish development) was limited to ancient Cantabria, Old Castile north of Burgos, and that the change was not general in Spain at all as Ascoli and others had supposed. From Cantabria it spread south and with the reconquest it invaded Central and Southern Spain. Later the aspirate h- was lost almost entirely in Castile but remained in the south.. Aragonese, Leonese, Portuguese and Catalonian did not participate in the development and kept Latin f- intact. The Cantabrian change Menéndez Pidal believes was a straight substitution because the Cantabrians could not pronounce f-. He assumes the same development for the Cantabrian territory in Gascony, where Latin f-> h- also. The reviewer believes that this chapter of Menéndez Pidal's great work is absolutely final and conclusive with respect to the entire history of the problem after the early appearance of h- < ƒin Cantabria, but is not quite satisfied with the theory of an immediate substitution due to Cantabrian speech habits. The problem of source might be studied also from the view point of Vulgar Latin and Italic dialect sources. Historical phonetics reveals much evidence in favor of a bilabial aspirate f- in Vulgar Latin, especially in certain regions. Latin f- represents the Indo-European voiced aspirates bh, dh, that became in primitive Italic ph, th, then f.5 The Indo-European gh-, however, became generally h-. In Oscan this h- was very weak and often omitted in writing exactly as Latin f-> h- in early Castilian and then became silent: Errant, Ferrant, Hormaza, Ormaza, (Origenes, p. 231). Furthermore, in Latin f- and h- were often interchanged Buck, Grammar of Oscan and Umbrian, Boston, 1904, §§79-81. 'Sommer, Lateinische laut- und formenlehre, §§105-8. 6 Buck, op. cit. §149. as in the early Castilian texts." The Castilian development then is only an extension of a well-known phenomenon in Latin in so far as the frequency of interchange is concerned. As for the early fall of the h- <f- it is only an extension of the parallel development of Oscan h-Indo-European gh-. In §43 we have further evidence for the linguistic unity of Spain up to the end of the Visigothic period. The initial groups cl-, fl, pl, remain unchanged in the entire Hispanic territory up to the end of the tenth century. The Castilian palatalization begins in the eleventh century. It may have begun earlier, Menéndez Pidal believes, but if so it was apparently considered too plebeian and avoided in writing just as -illo <-iello < -ěllu had been during the eleventh and twelfth centuries. In §§45-6 the Vulgar Latin forms with voiced intervocalic consonants and posttonic and pretonic vowels retained found in Leon in the tenth century are studied again. Many of these are studied in §§32-4. In §88 the problem is treated again. We are face to face with a Leonese Vulgar Latin preserved since Visigothic times thanks to the presence and influence of Mozarabic scribes. The development of medietate in Spanish has a chapter all by itself, §48. Meatad predominates in Castile, meetad in Leon. Meitad is frequent in Castile and Aragon, and the modern Castilian mitad is apparently Aragonese. The presence of t is explained as a learned influence. In §51 we have a fascinating study of the development of the groups -ct-, -lt-, to -it- in Leon, Aragon, and Navarre; and to -ič-> č- in Castile. The extreme development was reached in Castile in the eleventh century because in the Glosas Emilianenses and Silenses -it- is still found, feito, muito. Menéndez Pidal does not consider these as really Castilian forms but rather as Castilian archaisms, also preserved in Toledo as late as the twelfth century. In Mozarabic our author finds evidence for the more archaic -ht- from -ct-, the Oscan-Umbrian stage of Italic and Latin -ct, and also -it-, the Umbrian stage of secondary -ct-. The general Romance development is apparently parallel to the Oscan 7 Sommer, op. cit. §114. It may be that the original Latin bilabial ƒ remained in some of the Romance territory despite the objections of some philologists. See Grandgent, Vulgar Latin, §320. Lindsay, The Latin Language 99, believes that Latin f was at some time bilabial. He is wrong, of course, when he states that ƒ is bilabial in modern Spanish. It is true only for certain dialects. See my Studies in New Mexican Spanish, I §100. Umbrian and may be related to it. If in Mozarabic one still finds the Oscan stage -ht-<-ct- Castilian must have developed with marvellous rapidity to reach the -č- <-it- stage in the eleventh century. The Castilian developments mb > mm > m, nd > nn > n, ld > ll > l, lt > ld, discussed in §§52-4, Menéndez Pidal believes to be of Oscan origin. He finds that these Castilian developments irradiate from the territory around Huesca, < Ŏsca, the city of the Oscans founded by Sertorius. This region is east of and in general an extension of the original Castilian Cantabria whence f-> h- irradiates, and for that reason the reviewer believes that the evolution of that phenomenon may be also of Oscan source. The Oscan source of mb > mm > m, nd > nn > n seems definitely established. The comparative linguistic map of page 304 leaves no room for doubt. In §58 we have another chapter of the utmost importance for Romance linguistics, the problem of the development of secondary Romance groups, such as m'n > mr > mbr in Castilian. This is of course not limited to Romance philology. As for Spanish the general problem is how it happens that original Latin mn becomes nn > ñ in Castilian while secondary m'n becomes mr > mbr. Not counting the common form omne < homine, mbr prevails in Castile in the tenth and eleventh centuries (93% of all cases). The opinion expressed for the difference in the development of primary mn > ñ and secondary m'n > mr > mbr in Castile, namely, that in the original Latin mn, as in original pt, the tendency to assimilation, mn > nn > n, pt> tt > t, is explained by the weak articulation of consonants long juxtaposed, whereas in the case of secondary m'n there is recent vocalic syncope with necessary emphatic and clear pronunciation of the two constituent elements and. dissimilation favored, m'n > mr > mbr, is certainly more attractive than that of Millardet who thought of an anti-etymological and unphonetic syllabication, no-mne;s but it is not absolutely convincing for Leon and other Romance languages and dialects, where secondary m'n, for example, develops through assimilation to m, and it is precisely in Leon also that the hiatus between the two consonants was more pronounced if we are to believe that the numerous tenth century forms with the posttonic and pretonic vowels preserved are really examples of popular pronunciation. In §§61-80 Menéndez Pidal studies the morphological and syntactical problems of the early Spanish texts. It is not possible to review these studies in the present article. The new forms, such as tan mientre < 8 * Linguistique et dialectologie romanes, Montpellier, 1923, pp. 293–6. tam interim, algo[n]dre < aliunde, ad abiesas < ad aversa, yestra, gestra extra, adta < Arabic hatta, etc., are of fascinating interest to Romance philology. The chapters devoted to lexicology, §§81-5, and the final historical, geographical and linguistic studies of the various regions and epochs of Spain from the eighth to the tenth centuries deserve special and separate reviews which we cannot undertake now. Origenes del español is not only a work of capital importance for the study of Romance philology and epoch-making for Spanish historical grammar. It may be justly called one of the outstanding contributions made to the science of general linguistics during the last twenty-five years. It is to be hoped that Menéndez Pidal may have the time and health to finish soon his monumental Historia de la lengua española for which the present work is, as the author tells us in the preface, only an introduction. AURELIO M. ESPINOSA Le Mystère du Langage; les sons primitifs et leurs évolutions. Pp. 102. By CHARLES CALLET. Paris: Maisonneuve Frères, 1926. The author claims to have examined a great number of languages, and says (p. 6): 'J'ai pu remonter jusqu'à la bouche même de l'Hominien, retrouver les cris qu'il jetait, alors qu'il n'était encore qu'un animal; déterminer le premier sens de ses cris, lorsqu'il se dégagea de l'animalité et que les cris devinrent des mots; déterminer la raison profonde de leurs dérivations, alors, j'ai vu couler, lumineux, tragique, fatal, le large fleuve des vocables'. So (pp. 8-9) we find that all language starts from nasalized snarling, bellowing, whistling, and guttural snarling, characterized by the sounds gny ny, m, sy, r k gre respectively. For example (pp. 14-15), ny = tooth explains na = light, because of the glittering whiteness of the tooth. Whereupon we receive a truly large fleuve de vocables, continuing with but slight islands of explanatory text, until we reach the concluding chapter of two pages. Examples are drawn from indigenous languages of Canada and of Peru, from Sudanese and Senegambian dialects, from Annamite and Malay, from Chinese and Japanese, Magyar, Hebrew, Assyrian, etc. Of languages familiar to the reviewer, there are Sanskrit, Greek, and Latin: in which the words appear freely misspelled. Three successive examples of Latin on p. 45 are 'Silio avoir soif, sabaya = boisson, sabazies = fêtes de Bacchus'. And when on p. 29 he finds: 'Nava = noeuf (sic!), nouveau (grec neos; islandais, nua). Ces mots se rapportent au bourgeon qui = |