Summary.

39. The dialects may be divided into five classes in point of rhotacism, though perhaps to little purpose, since a characteristic of this kind argues very little by its presence or absence for the affinity of any two dialects in other respects. Indeed the whole argument from geography rests on the 'chain' as opposed to the 'tree' theory.

SHEWING THE PREVALENCE OF

RHOTACISM IN THE ITALIC DIALECTS

AS EVIDENCED BY TRADITION, INSCRIPTIONS, AND LOCAL NAMES.

I.

II.

Names in brackets are included under the name they follow.

III. Dialects in which there is no evidence but that of geographical contiguity.

IV. Dialect in which Rhotacism was probably absent.

V.

Marrucinian.

Dialects in which Rhotacism was certainly absent.

1.

Pelignian.

2. Sabine.

3. Oscan (Bantian).

The result is embodied in Mr Heawood's map. Names whose form is of importance are printed in ordinary type: those in Italics shew the distribution of the dialects as denoting the places where inscriptions have been found: those in capitals have no reference to the argument. Modern names are enclosed in brackets. The ground plan of the tribal divisions is enlarged and slightly modified from those given by Mommsen (Unterit. Diall.) and Droysen (Historische Handatlas).

tabt

V. S between vowels in Latin.

A. Introductory.

Latin.

40. The exceptions to the change of s to r between vowels in Latin have always been recognised but their Aspect of the number seems comparatively small, and at first question in sight unlikely to suggest any substantial modification of the principle. So far as I know no endeavour has hitherto been made to limit the rule by any conditions which would account for the cases in which it is apparently suspended. The theory maintained in what follows is not without difficulties (notably virus āra), and without the strong confirmatory evidence of the other dialects could scarcely be called established. It is simply an attempt to discover some characteristic difference common to the majority of the forms to be explained, in lieu of the separate hypotheses hitherto advanced to account for them singly. The rule was originally suggested by the changes in Teutonic and Sanskrit, but it did not reach its Preliminary present form until it had been tested by all the ex- considerations in support of amples I cod find from any available source, such the theory. as the indices of Varro and Festus. When the tale External: of evidence was complete the theory seemed to me (1) Accent. to be greatly confirmed (1) by the exact coincidence (2) Phonetics. of the corollaries which followed from it as to the change of the Latin Accent with what was already known independently on the subject, and (2) by the greater ease with which the result seemed explicable from a physiological standpoint (v. supr. § 6-8). Finally my faith in the rule as it had so far been developed was strengthened by (3) the fact that of Internal: the various schemes by which from time to time it (3) Economy seemed possible to reduce the mass of facts to of material.