DATA AND NOTES. 371 



Saparua, Teluti : main, soft. Matabello, Gah : maluis, id. Amblaw : 

 maloh, id. Batumerah: mahita, id. Malagasy: lemi, soft- 

 ness, meekness; malemi, soft, meek, gentle. Malay: lamah, 

 soft, flexible, weak. Java: lamas, id. 

 Arabic: haluma, halim', to be gentle, weak. 



The stem is malum or malumu. This is in form a conditional of lumu, 

 a primitive not yet as such identified. 



In the Polynesian we have the utmost abrasion in main. This may per- 

 haps be identified with Lambell mala. I have associated herewith Lamassa 

 manlu; we are not yet sufficiently acquainted with this recently reported 

 language to know whether n is such an infix as is common in the not distant 

 Indonesia, or if man is a local form of the common wa-conditional. 



The malum form appears in Efate and Malekula. Vaturanga maluka is 

 not properly in this family ; it forms a small group with Eromanga molok- 

 loku, Sesake manukunuku, Mota manoga, and Motu manoka. On the m-n 

 mutation in Tangoan Santo malum see note 312. The Duke of York galom 

 may not be malum, but there can be no doubting the identity of lorn. 



The subduplicated malumlum occupies an interesting position in the New 

 Hebrides and is illuminative of the manner in which loan material is broken 

 in foruse byan alien race. Itwillmore convenientlybe studied by dissection. 

 The ma-conditional retains its a in Mae wo, Merlav, Malo, Mota. It becomes 

 me in Vuras, Santo, Lo and Volow. Mosin and Norbarbar have altered it 

 to mo. Pak, Sasar, Motlav, and Alo Teqel have allowed it to degenerate 

 to mu. If we regard this last as an attraction to the vowel of the lum 

 element we shall find the same principle of attraction operative in Volow 

 wielemwlemw, but not in Lo melunglung. The lum-stem remains unaltered 

 over the greater part of this subdistrict ; in Lo and Alo Teqel it becomes 

 lung; in Volow it not only changes to the difficult mw, but accompanies it 

 with a vowel change to e. 



Aneityum mulmul offers a problem. Codrington (Melanesian Languages 

 91) suggests the probability that it is metathetic for lumlum; metathesis is 

 rarely employed in Aneityum, but see pula (284) to shine Aneityum laav, 

 lav, and it would be the solitary instance in which we have identified the 

 primal stem lum. On the other hand Aneityum has taken such liberties 

 with its Polynesian loan material that if we regard mulmul as a degraded 

 mulumulu the foregoing note as to the impracticability of Efate malmal 

 would have far less application. The duplication anomalies of Duke of 

 York m alum alum and Laur malmalungana are less considerable when we 

 record that the remote languages of the eastern portal have developed their 

 duplication mechanics along lines quite other than those which I have 

 established for Polynesia. 



In Indonesia the best identification, despite wide vowel diversity, is the 

 Malagasy. Inverting the apparently metathetic Java and Malay forms we 

 find stems in s, malas, malah, akin to Amblaw maloh, probably to Matabello 

 and Gah maluis, and by a frequent change akin to Batumerah maluta. For 

 this reason the malu of Saparua and Teluti, despite its present identity with 

 Polynesian malu, probably is an abraded mains. 



In the Arabic haluma we should have a close resemblance if it were 

 possible to establish the identity of ha as a conditional and homogenetic 

 with ma. 



