250 Proceedings of the Royal Society of Edinburgh. [Sess. 
been any variations in the twist. Twist distributes itself along the length 
of the thread according to the number of fibres present in each section. 
[The precise relation which connects these quantities will be discussed 
in a future paper on the distribution of twist in a thread.] It is 
sufficient for our present purpose to state that in drawing the sliver 
or roving, the fibres do not as a rule distribute themselves uniformly 
along the thread. When twist is subsequently put into the thread during 
the spinning operation, it runs to an excessive degree in the sections con- 
taining the smaller numbers of fibres. When the sliver is very uneven or 
“ pointed,” this action of the twist forms local hard, small parts in the 
thread, which are technically called “ twits.” However, the fact must never 
be overlooked that these are always present in the best-spun yarns, but to 
a smaller extent in number and degree. We have then at least two causes 
of the phenomenon observed by the Manchester investigators, viz. : higher 
degree of twist and lower number of fibres, or vice versa ; lower degree 
of twist and higher number of fibres. The strength increases with the 
twist, and decreases as the number of fibres decreases, but not necessarily 
in the same proportion. Indeed, it is possible that the twitty parts are 
stronger than the average, but only if the “ twit ” or unevenness is not 
excessive ; i.e. the strengthening of the thread due to twist may be greater 
than the weakening due to decrease in the number of fibres.* But the 
point at issue remains, that there were at least two causes of the pheno- 
menon, and unless these were isolated and their effects determined separately, 
any assumption as to their single effects on the strength of a thread is 
unwarranted. The observations taken were not due to the effect of twist 
alone, but to the superposition of the effects of twist and at least one other 
cause operative at the same time. 
(2) The second flaw in their argument appears to be that the yarn that 
was tested for twist was not the same yarn that was tested for strength. 
Fifty-eight threads were tested for twist, after testing ninety-nine threads 
for strength. The probable reason for this procedure might be that it was 
not possible to vary the twist on the machine during the test ; therefore 
two operations would have been necessary on different machines for the 
same thread. The Schopper testing machine, although the best for the 
main purpose that the experimenters had in view, is not the best for this 
side question of the relation between twist and strength of the thread. 
This difficulty, however, put them in the position to some extent of 
assuming what they had to prove. 
(3) Thirdly, the limits of variation of twist were too close together to 
* This question will be more fully discussed on p. 259. 
