MELANESIAN ANNOTATIONS ON THE VOCABULARY. 33 
and the New Hebrides if it is to be considered common with the Ndeni 
and Nifilole forms. Uni exhibits recognizable occurences of the com- 
mon migrant material, but this lies outside that class. 
vegara (57). Kiviri and Kubiri are separated by no great distance 
across Collingwood Bay on the north shore of Papua; the two lan- 
guages are in general similar; the mutation n—ng is common in this 
area. In Ray’s material this is the only instance of its occurrence 
between Kubiri and Kiviri. On the other hand, I note but a single 
instance in which g is common to the two languages. 
kelama (59). ‘The form keyama is either subdialectic or individual 
in Mukawa, for we find the kelama form as well. We have several 
instances of this l-y variety within Mukawa. 
rogona (60). We shall regard this as the type form, rogoma the 
derivative, for in a later instance (ain 4) we shall find ani in the general 
stem and am as a Taupota secondary form. 
The remaining forms in the list occur but once each; not one seems 
associable with any other. 
STEM ai. 
This is numerically the most frequent by far in New Guinea and 
Melanesia. We find it as a nude stem; we also find it with a palatal 
prefix. As to the latter point we may not speak definitely as to whether 
it is anassumption upon thenude stemor the word has undergonefrontal 
abrasion from a primal stem kaiinto the form ai. In the recent variety 
of speech in the Pacific we have abundant evidence of the disappearance 
of k as objectionable, followed by the resumption of that consonant as 
coming once more into fashion; but in this resumption it is not restored 
to the place which comparison with unaffected languages shows it should 
fill, for it reappears as atcontamination. ‘To assume a primal ai stem 
reasonably accords with my theory of the growth of these languages 
by the application of consonantal modulants. 
aie is reported by Neuhauss from Sissano, where Friederici finds the 
simpler ai. To the latter we owe its ascription to the eastern Bis- 
marck Archipelago, geographically remote. With it we may compare 
32-34, in which awe is found in the western Bismarck Archipelago 
and kaiwe and kaiwa on islands at the southeast promontory of New 
Guinea. In the Lamassa we may identify ai with prefix of i and from 
this we may pass to ia. Assuming the consociation of ei with ai, and 
Yakomul is in the region on the north coast of New Guinea where ai 
predominates, we note the corresponding series ei-iei-ie (3, 5, 6). 
We shall base no argument on these two series, for each rests upon a 
single instance in each of its members, but we enter them upon the 
record for such value as they may possess. Ray makes an obscure 
note upon the series of words signifying the bow, “‘its stick (7. ¢., of 
arrow).” ‘This note applies to Awalama aina, Taupota pidu-aina, 
