MELANESIAN ANNOTATIONS ON THE VOCABULARY. 67 
guages in which the sonant mute is so imperfectly acquired as to call 
for the help of the preface of the nasal of its own series, mb, in order 
to secure its pronunciation at all. In a smaller group we find the 
spirants, surd f in a compact language group in Papua, surd v once 
occurring in a region far remote. Even the semivowel proximate to 
the labial tract, w, is once met with in the variety of this vocable; 
other stems in which it is employed with some greater frequency tend 
to corroborate this instance. The stem is almost uniformly dupli- 
cated or in a condition reduced from duplication by abrasion of the 
final vowel. ‘The nude stem appears but. once in 40 mbe; in com- 
paction with other stems we may trace it in 33 kamb4, 39 uleulebe, 61 
kapeu, and 62 gopu. The vowels exhibit considerable variety, as shown 
in this list of the characteristic vowel of the monosyllabic forms and 
the combination of the two vowels in the completely duplicated forms: 
= gana?) 34,40. a-—a 42, 43. 
S-e i-5, 16-22, 23, 24-30, 38, 55-56, 58. | a-e 14, 4I. 
e€-a 35, 36. a-i 12, 13. 
e-i 37. O-O 47-52. 
i 10 o-€ 59. 
i-i 8, 9, 11, 45, 46, 57. 
In 23, 33, and 34 the vowel is recorded by a German authority as 4. 
Inasmuch as this recorder has not been generally confirmed by the 
succeeding explorations of Friederici in the same field, I attach no 
great weight to this umlaut, and for practical purposes have included 
these forms in the e classes. In the ‘Polynesian Wanderings,”’ 
loc. cit., I have extracted from languages about the Gulf of Papua a 
comparative series of forms which at several points bear close resem- 
blances to the pepe series, this series carrying the signification of the 
wing. The association in sense is so noteworthy that it would not be 
improper to suspect a community of source in some archetype stem 
of such scope of meaning as to admit both the vocables. 
In his argute elucidation of 49 nal bobd Friederici traces out a most 
interesting series of compactions of the bird-butterfly sense. Unfor- 
tunately we lack data upon which to resolve the other compactions 
of this series, yet the recurrence of certain types can not be without 
significance. In these lists the position of the hyphen denotes the 
place in the compacted form occupied by the extraneous element, 
whether initial or final. We segregate a group in which some commu- 
nity may be imagined: 39 ule-, 2 le-, 29 —ula, 3 —le, 38 —lo, 14 —lug. 
Another group is found as initial—the series seems well established: 
46 kara—, 9 ndra—, 30 ara-, 42 kori-, 54 kili-; once only is this found 
as final, 52 —koro. A final series appears in 28 —ruka, 25 —roho, 
26 -ru, 27 and 61 -u. Another is 12 —wa, 4 —pwa, 5, 31, and 59 —a. 
Yet another is 50 —be, 24 —bi, 13 —pi, 11 —p. ‘Three forms are repre- 
sented singly—i19 kau-, 15 —k, 51 —di. 
In closing, a few notes should be made upon certain of these forms. 
Tanna has particularly deformed its borrowings from migrant races; 
