58 
circuit while the thread is at its standard size ; for under such condi- 
tions the lever is so held by the thread that the contact at a is kept 
open. As soon, however, as the thread diminishes in size the lever 
recedes, the contact at a is closed, and the current passing through the 
magnet of the feed movement W causes the attraction of its armature 
and the release of the detent holding the cam in place. Upon this occur- 
ring the magazine is advanced one step and brings a new cocoon fila- 
ment into the path of the hook on the filament attaching device, which 
catching it up twists it around the running thread and, with the help 
of its natural gum, attaches it firmly thereto, at the same time cutting 
off the loose end. The rotation of the cam is so timed that its detent 
will not arrive at the stop on the armature until the new filament has 
reached the controlling drums and had its effect upon the position of 
the control lever. In the reeling of fine sizes the addition of one fila- 
ment will generally be found sufficient to bring the thread to its normal 
size, though it is less apt to be so with larger sizes. In any case, how- 
ever, if, when the rotation of the cam is completed, the electric circuit 
still remains closed the action of the feed movement is repeated and con- 
tinued until the thread is again brought to the normal size. 
Owing to the irregularities in a thread of raw silk it is impossible to 
obtain any measure of its size by means of a caliper or even, with any 
degree of ease, by a microscopical examination. Merchants are there- 
fore obliged to content themselves by approximating its size in the fol- 
lowing manner : They measure off upon a suitable real a skein of a 
given length (476 meters) and obtain its weight in the terms of an arbi- 
trary unit called the denier. If such a sample skein, for instance, is 
found to weigh ten deniers it is called a u ten-denier silk.” Now it is 
found that the exterior thread of a cocoon of the yellow Milanese races 
has a value of about two and a half deniers, so that it takes four such 
new cocoons to make a thread of ten deniers. When these cocoons are 
half unwound the size of the thread formed from them would be but about 
eight deniers. Now, in order to augment the thread and bring it to the 
normal size we are obliged to add another cocoon which, with its new 
thread, would increase the combined thread to ten and one-half deniers, 
and it will be seen that from cocoons of this race it is impossible to 
augment the thread by smaller increments than that mentioned. For 
this reason no attempt is made to produce an absolutely regular thread 
of silk, but reelers are content if the variation from the desired mean 
does not exceed two deniers in each direction. In hand-reeling, where 
the regularity of the thread depends eutirely upon the ability of the 
reeler to estimate its present size and to add a new filament at the 
proper time, only the most expert operatives are able to make silk with- 
in the limits named. In the automatic reel, however, all this is taken 
out of the hands of the operative and the indication of the need of a new 
thread is made by the delicate serigraphic measuring device of the con- 
trol movement. Its delicacy is such that when working under good 
