^{page 737 note 1} Vowel length is the phonetic feature involved; its significance is discussed later.

^{page 737 note 2} A. Capell, A New Fijian Dictionary.B. frequently disagrees with the lengths marked.

^{page 737 note 3} C. Maxwell Churchward, A New Fijian Grammar.

^{page 737 note 4} f, commonly pronounced as v,is used in the Tongan word faiwa

^{page 739 note 1} B. will not, however, accept a fully–voiced b without a preceding m.

^{page 741 note 1} The only exception I can find is uwea,pronounced u′wεe, a kind of fish-trap. This is taken from the native name of Wallis Island, and is a foreign word (spelt by the French Ouvéa).

^{page 741 note 2} I do not, however, agree with Capell′s use of the length-mark on the first a. It ia unnecessary, and to some extent misleading, since there can be no distinction by length at that point. See below.

^{page 741 note 3} B. gives ie, ege, evorovoro, evuevu, iota(Eng. yacht),for Capell′s yē, yege, yevorovoro, yevuyevu, yota.

^{page 742 note 1} It is not to be understood that –ya, –iare the only suffixes occurring after o and u.Cf. davo–ca, bulu–ta, davo–ci, bulu–ti.

^{page 742 note 2} a appears, apparently as a suffix, after o and uas well as other vowels in certain words (often containing a reduplicated element) that are usually classed as adjectives, e.g. ulouloaand dravudravua,as well as savasavaa.The attempt to apply the categories of the Latin parts of speech to Fijian is of doubtful validity, and the whole question needs examination. There does, however, seem to be prima facie evidence here of a different category from that containing the words mentioned in the text; at least of a diiferent suffix.

^{page 742 note 3} See H. Kern, De Fidjitaal.

^{page 742 note 4} Many words have, in certain grammatical relations, a suffix consisting of one of the following: –a, –va,–to, –ra, –ka, –ma, –na, –ga, –ca, –ya;or of –vaka, –taka, –raka, –kaka, –maka, –naka, –laka, –caka.These have been called transitive suffixes. When the other word in the relation is a proper noun or pronoun, the suffix has the vowel»instead of o. See Churchward, op. cit. Kern (Fidjitaal;see also S. H. Ray, Melanesian Island Languages)gives reasons on comparative grounds for supposing that the suffix consonant is, in some cases, the final consonant of the Indonesian root. In other cases, as he also shows, it is not. It does not seem that anything in Fijian indicates that the consonant does not belong to the suffix, though the large number of forms serving apparently the same function suggests a problem.

^{page 743 note 1} Three unusual cases occur: doonuya(or donuya),connected with donu; leevea(or levea),connected with leve;and draavuya(no dravuya),connected with draw,.It is possible that there may be a form of partial reduplication in these—dodonuand dravudravuoccur.

^{page 743 note 2} It may be noted, however, that B. rejects as unknown to him a considerable number of the three–syllable forms given by Capell.

^{page 744 note 1} Others, dissyllabic suffixes; the result being a three–syllable group plus a two–syllable group, e.g. maqosa–taka.

^{page 744 note 2} There is one anomalous case: the ε often used at the end of questions (and not to be confused with the particle e quoted above). It seems to mean something like “ really ? ”, and may be used alone, in which case it is ε: ? In such a case as a: ′ncnaε ? “Is it his? ”, it is short and is not treated as a separate syllable in the intonation.

^{page 745 note 1} It should be added that the forms noodaru,etc., occur only when the pronoun is used predicatively. Thus there may be either ′a:nc′ndaΤU or, a: ′nc:′ndaΤu “ it is ours ”, but only nc′ndaΤu ′βalε “ our house ”.

^{page 747 note 1} Numbers in parentheses refer to the examples.

^{page 749 note 1} tiko, sign of continuous aspect.

^{page 749 note 2} era aa.