MELANESIAN ANNOTATIONS ON THE VOCABULARY. 39 
Here we have exactly the series of consonantal modulants which we 
saw employed in the first item of these notes. 
The nude stem ani (18) is by far the most frequent type in Torres 
Straits; in fact these records identify it in a single occurrence outside 
this region, namely, in Ninigo of the northern Bismarck Archipelago. 
Herewith we include Mekeo angi (19) with the note that n—ng mutation 
is frequent in that speech. The demolition form an may be the 
result of the final abrasion applied to ani or ana, and the same uncer- 
tainty persists through the yet more fragmentary form a. In Barriai 
ean (20) and in Nufoor aan (21) we note the same sort of prefix as in 
Dobu e’ai (2). In accordance with the simple n-ng mutation we 
may include Jabim eng and ing as variants of an, this being the more 
probable since the material includes so large a proportion of ani material 
and lacks any other stem form from which this might derive. 
The ani stem in New Guinea appears both on the south coast and 
at the east; kani (6) is restricted to the east promontory and then 
leaps to the New Hebrides and to Viti. The first abrasion form kan 
appears in the Bismarck Archipelego. In kanega (8) of Tagula we 
find a suffix for which we lack explicative material; we note that 
suffixes appear in gania (11), gad (17), and in anan (24). 
Our gani series is represented from 9 to 17 and in the simplicity of 
its ordering calls for little more than inspection. Duplication is 
present in 12, 15, and 16, and in the hana seriesin 28. Arag nganngan 
(16) lies within the n—-ng mutation and is unmistakable. Nada gad 
(17) is a doubtful form; in the preceding paragraph I have included it 
with the suffixed forms as produced from a final abrasion type ga. 
Such a ga would fall in series with a which we have found to exist; it 
would then enlarge the series to such an extent that we might feel 
justified in interpolation of a theoretical ka for the establishment of 
the Nengone kaka (40). Yet since the whole question of a suffix rests 
on scant and uncomprehended material we may not give this sug- 
gestion whole support. It is far more probable that this gad is asso- 
ciated with the Polynesian kati stem as meaning to bite; in fact, Ray’s 
material collated in that sense seems to establish this beyond perad- 
venture. 
The hani series, quite typical in the treatment of nude stem with 
consonantal modulant, is very brief in this material. Keapara yields 
us the type form hani (26); the lani (27) of Kelana in western Bismarck 
Archipelago is anomalous, for the type of consonant modulant in these 
stems lies in the palatal series and here we have a lingual liquid. This 
removes it from the possibility of mutation; but the liquid is, in speech 
evolution toward consonant possibility, an earlier acquisition, therefore 
we may look upon lani as a treatment of ani in the process of partic- 
ularization more primitive than the palatal series of which kani is 
the type. 
