90 W. H. WARREN. 
as small as one ten thousandth of an inch, and for the determin- 
ation of the true elastic limit the error in the measurements must 
not exceed one hundred thousandth of an inch. 
The apparatus which has been hitherto in use in the Engineer- 
ing Laboratory for the measurement of small strains consists of 
various arrangements of levers or micrometers. The most delicate 
of these are, a, the Lever Extensometer designed by Prof. Kennedy; 
6, the Richle-Yale Extensometer. 
Prof. Kennedy’s Extensometer consists of a light frame attached 
to the test piece, and carrying a light lever multiplying the strain 
a hundred times, and giving the mean strain produced on each 
side of the test piece. The scale is divided into tenths of an inch, 
but it is possible to record one-tenth of these divisions, in which 
case the readings are taken to one ten thousandth part of an inch. 
The Richle-Yale Extensometer consists of a light frame attached 
to the test piece, and carrying two screw-micrometers which 
measure the extension or compression of the bar on each side to 
one ten thousandth part of aninch. An electric battery and bell 
are attached to enable contacts to be made with the micrometers, 
with greater accuracy. 
Professor Martens’ Mirror Apparatus is far more delicate than 
either of the foregoing, and has recently been made for the 
Engineering Laboratory by Mr. Edward Béhme, instrument 
maker to the Royal Mechanical Technical Experimental Station 
Charlottenburg, Berlin. It is represented in the accompanying 
sketches, Fig. 1-4, and consists of two small prisms &* which 
are held in firm contact with the test piece and the distance 
pieces dd by means of a steel wire spring, the action of which is 
indicated by the arrows ss. Each prism is provided with a stem 
a, which carries a small mirror m held in the frame ff rotating 
freely about the stem, and is held in position by means of a spring ¢- 
At b is a capstan screw for the adjustment of the mirror which is 
held against its point by a small spring not shown in the sketch. 
At a definite distance from the test piece are two stands side by 
