252 THE POLYNESIAN WANDERINGS. 



Arag, Ulawa, Saa, Fagani, Bululaha, Hula, Keapara, Galoma : pepe, 

 butterfly. Epi: lepepe, id. Moanus: ndraplpi, id. New 

 Georgia: pepele, id. Omba, Maewo, Wango, Vaturanga, Buka, 

 Baravon, New Britain, Solomon Islands, Rubi, Suau, Sariba: 

 bebe, id. Laur: bdbd, id. Mugula, Tubetube, Tagula, 



Murua: bebi, id. Savo: bebeula, id. Motu, Sinaugoro: 



kaubebe, id. Panaieti, Misima: bebebi, id. Pokau: ebebelo, id. 

 Roro: peropero, id. Nada, Kiriwina: beba, id. Boniki: 



bebabeba, id. Mukawa, Raqa: arabebemta, id. Kwagila: 

 karabibim, id. Taupota, Wedau: bebeu, id. Awalama: 



kapeu, id. Tavara: gw/w, id. Dobu: pepega, id. Mekeo: 

 /e/e, id. Kiviri, Oiun: /e/e&, id. Lakon, Pak, Sasar, Alo 

 Teqel: pep, id. Lo: pip, id. Norbarbar: peb, id. Volow, 

 Merlav, Gog, Motlav: beb, id. Lamassa: 6aw, id. Lambell: 

 'hamb'd, id. Mota: pepe, rupe, id. Nggela: uleulebe, id. 

 Buka: 6awe, id. Brierly Island: bebi, a moth. Malo: vebe, 

 a butterfly. Baki: bembe, id. Tanna: paubauuk, id. 

 Morella: pepeue, butterfly. Amboyna: pepeul, id. 

 Hebrew : 'w/>, '«6'e6, to flutter. 

 The identification is so complete that we need no more than simple 

 inspection until we reach the forms which are either more or less than pepe. 

 The pep, by abrasion of the final vowel, finds itself in Rotuma, Lakon, 

 Pak, Sasar, Alo Teqel, Lo, Norbarbar, Volow, Merlav, Gog, Motlav. Still 

 further abrasion, or what amounts to the same thing, the unduplicated stem 

 pe, does not, with perhaps a single exception, appear unsupported ; but as a 

 composition member it is found in Lambell 'hambd, Mota rupe, and Nggela 

 uleulebe. Our possible exception is Lamassa bam. Laur, Lamassa, and 

 Lambell we have through a single channel, very German and the most com- 

 petent piece of ethnography I know. From this we see that bd is close to 

 be. In Lambell the bd is compounded ; in Lamassa, but a few miles away, 

 bam may be regarded as but bd nasalized, as is sometimes the Melanesian 

 case; or the m may be a mutant from p, as we shall discuss in note 207. 

 Variants of pepe are easily traced in Malo vebe, Baki bembe; and Tanna pau- 

 bauuk, while remote from the standard, as Tanna lies remote from the free 

 track of migration, seems struggling to preserve the same stem. In Indo- 

 nesia the pepe stem is unmistakable in Morella and Amboyna. 



We now turn to the pepe composites. Lambell 'hambd we have already 

 noted. Not by any means identifiable with 7*aw,but associable because at 

 the northern verge, we note Motu kaubebi and Moanus ndraplpi. There is 

 a liquid component which so frequently associates itself with pepe as to 

 attract notice. Its simplest form is le, initial in Epi lepepe, final in New 

 Georgia pepele. There is a stronger form, yet I incline to regard it identical : 

 ula final in Savo bebeula, Amboyna pepeul, Morella pepeue; initial as tde in 

 Nggela uleulebe. This may appear in oriori butterfly in the Lakatoi language. 

 This consistent reducibility to elemental pe, and Buka bawe is so reduc- 

 ible as well, removes all possibility of identification with the Hebrew word 

 offered. 



We should not pass unnoticed a group of forms which Dr. Ray has col- 

 lected on the New Guinea shore of Torres Straits. The meaning is wing, 



