42 BUEEAU OF AMEEICAN ETHNOLOGY [bull. 40 



2. Somewhat less transparent is the former existence of a -w after 

 consonants. The f ollo\\dng examples have been found in the material 

 at disposal: 



Idl she twined basket ( = Halw) ; cf . Id'^lwa'^n I twine it (that -w 



really belongs to the stem is shown by the forms ld°-wa'n 



I shall twine it ; leUxi twine it for me !) 

 Tclel basket bucket ( = *llc!elw); cf. Jclelwl'^ her bucket 

 ¥al penis ( = *¥alw) ; cf . Valwl''^ his penis. 

 sgele¥ (=*sgelel^w) he keeps shouting; cf. sgelewaY you shout, 



sgelwa'We^ I shall keep shouting 

 alsgdWa^ ( =*sgdlw]c'a^) I turned my head to one side to look at 



him; cf. alsgdHivi'n I shall turn my head to look at him 

 alsgelelxi (=*sgelehvxi) he keeps turning his head to one side to 



look at me; cf. alsgald^iwi'^n I keep turning my head to look 



at him, future alsgalwalwi'n 



This process, as further shown by cases Hke gal eat it! {=*galw), is 

 really a special case of the simphfication of double diphthongs (see 

 § 11), Perhaps such "dissimilated" cases as Id'^- and le^- (for Idu- 

 and leu-), see § 7, really belong here. 



Other consonants have doubtless dropped off under similar condi- 

 tions, but the internal evidence of such a phenomenon is not as 

 satisfactory as in the two cases listed. The loss of a final -n is probable 

 in such forms as ihegwe'hak''" he works, cf. iTiegwe' haV'^na^n i work, 

 and I'hegwe'TiaV'^nana'V we work. Certain verb-forms would be 

 satisfactorily explained as originally reduplicated like gwidiW", if we 

 could suppose the loss of certain final consonants : 



gini'^V he went somewheres ( = 'i*gin-i'-^Vn) 



gelgulu^F'^ he desired it (= 'i*-gul-u^-FH) 

 In the case of these examples, however, such a loss of consonants 

 is entirely hypothetical.^ 



§ 19. SIMPLIFICATION OF DOUBLE CONSONANTS 



Morphologically doubled consonants occur very frequently in Ta- 

 kelma, but phonetically such theoretic doublings are simplified into 

 single consonants; i. e., Tc'+g become V or g, and correspondingly 

 for other consonants. If one of the consonants is a fortis, the simpli- 

 fied result will be a fortis or aspirated surd with preceding catch, 

 according to the phonetic circumstances of the case. If one of the 



1 Many of the doubtful cases would perhaps be cleared up if material were available from the upper 

 dialect, as it shows final clusters that would not be tolerated in the dialect treated in this paper; e. g. 

 h'ii'iina'ks-t' relatives (cf. Takelma h'lvinaxde my kin). 



§ 19 



