PHYSICS: C. BARUS 
41 
secured by bolts etCi, as at u. The spring k draws the lever toward the front 
of the diagram, so that the blunt metal pin e suitably attached to the end of 
mm' may be kept in contact with the glass plate g to be tested.' 
The plate in order to be examined as to its degree of plane parallelism, 
must be capable of sliding up and down or right and left under standard 
conditions. To obtain these the stout bar G rigidly attached like / to the- base 
of the apparatus, has been provided, carrying three set screws h, h, h, the 
points of which lie in the same circumference about 120° apart. They there- 
fore constitute a kind of tripod against which the plate g is firmly pressed by 
the flat spring or clip rr and screw i. This method of mounting may be 
appropriately varied in accordance with the tests to be made on the plate 
its shape, etc. Similarly the set screws h, h, h, may be placed nearer to- 
gether or further apart in appropriate screw sockets, and finally the lever c 
may be lengthened or shortened at pleasure. The pin e remains in per- 
manent contact with the plate g in consequence of a wide circular hole in the 
clip rr; or e may clear rr, above or below it. 
If but one face of the plate g is to be tested, the system Ghrg must slide as 
a whole right and left, nearly parallel to the rays pq. In such a case every- 
thing will depend on the excellence of the slide carrying the system. I did 
not attempt to make such arrangements, as I had no need of data of this 
kind; but parts MM', NN', Fcmm' and Org were nevertheless mounted on 
heavy slides (lathe-bed fashion) for convenience in securing a variety of 
adjustments. 
In figure 3 the bar G has been reversed in position and the contact pin e 
now passes through a circular hole in G to be in contact with a lens g, for 
instance, kept pressed to the tripod screws h, h, h, in the same way as before. 
The latter should in general be much closer together than the figure shows. 
The instrument is now a spherometer. 
The experiments indicated that the mounting of the contact pin e to the 
extremity of the mirror mm' may be the occasion of annoyances. For on 
sliding g right and left or even up and down, the mirror mm' is liable to be 
flexed. In such a case the achromatic fringes rapidly lose sharpness, not to 
ipeak of the errors involved. I endeavored to avoid this by keeping the pin 
e out of contact with the plate g by a special lever (not shown), while g was 
being displaced and to test a number of successive contacts thereafter; but 
it is best to mount e on a separate rigid cross-piece parallel to mm' and firmly 
attached to c. In such a case no flexure of mm' can occur and the contacts 
may also be repeated at pleasure. Before each reading the bar G should be 
gently tapped. 
The achromatic fringes can be found only through the spectrum fringes. 
This is not usually difficult remembering that not only must the slit images in 
the spectrum be in contact throughout, but the two beams must be locally 
in contact on the mhror M'. Moreover the mirrors M' and N' must be 
equally thick and the silvered faces all turned towards the auxiliary mirror 
mm'. 
