118 SISSANO. 
we :find ourselves engaged uncertainly with a new formative element 
for which we have no clear explanation. This may be a formative 
syllable kar (gal) prefixed, in which case we should note that in the 
forms 32-34, occurring not only in the west but also in the north and 
in the east of Indonesia, we find such an element in kar-ang; but as these 
forms occur within the region of the formative infix, we must not 
neglect the possibility that they arise from the already prefaced golon, 
inferential from Subanu golo head, in the type form g-ar-olon, from 
which we may easily derive karaluni. Each suggestion is but a specu- 
lation; at present we want the data upon which to prosecute the 
research to a solution. 
52. lepi sago. 
37. lapia Ambon, Bahasa. 44. rambia Malay. 
38. labia Barée Toradja. 45. rombia Malay. 
39. labia Gorontalo. 46. rumbia Tontemboan. 
40. lepia Ambon, Bahasa. 47. humbia Sanguir. 
41. elpia Amblaw. 48. bia Gorontalo. 
42. lepial Ambon, Bahasa. 49. béa Bunda. 
43. ripial Ambon, Bahasa. 50. bi Vier Radja. 
In the forms 48-50 we find particular reference to the fecula and in 
certain cases the addition of a determinant element is required when 
extending this element to the tree which produces it; thus, in Gorontalo 
we find bia designating the fecula and la-bia the tree; in Buru bia and 
biapun respectively. The labia type is of such wide extent in the two 
western provinces as to constitute a special stem. A resemblance is 
noted to the general Polynesian la which exists independently in the 
sense of the branch or limb of a tree, which compacts with another weak 
stem in la-kau to designate a tree. ‘This is noted as a resemblance 
only in this connection through lack of intermediate data; in the 
“Polynesian Wanderings,” page 304, will be found a summation of the 
scattered occurrences of this word in Melanesia and in Indonesia. 
The Vier Radja bi seems to suggest an earlier type of the pia stem, 
for in languages which tolerate the open form exhibited by bi there 
must be the same tolerance for the open pia and therefore we can not 
see any reason to suspect that bi is an abrasion product. In the 
Melanesian series we find no evidence of the employment of bi inde- 
pendently, but in compaction with the la element we note its presence 
in the forms 10-21. ‘These are confined to New Guinea and the adja- 
cent Bismarck Archipelago, but the obvious conclusion as to source 
is complicated by the fact that certain of the occurrences are credited 
to the south of New Guinea and to peoples classed as Papuan, that 
is, non-Melanesian. In 41 elpia we find no difficulty in recognizing 
metathesis, in 42-43 the assumed final 1 is anomalous and receives 
no support elsewhere in the history of the stem. 
