THE AID OF THE ACHROMATIC FRINGES. 71 



and therefore E too large. The effect would be a marked increase of E with 

 the load such as occurs above. 



As a result of the residual difficulties enumerated, rods whose modulus 

 is of the order of io 12 can not be expected to show trustworthy behavior, 

 unless their diameters are less than a millimeter for a length of about 2 cm. 

 in a suitable protecting sheath. Rods whose moduli lie markedly below 

 io 11 may be used when the diameter is 3 or 4 mm. It is for these materials, 

 i. e., the large class of well-developed organic solids, that the apparatus 

 has been devised. From them, moreover, in view of the accentuated de- 

 formations, interesting information bearing on the elastics of bodies as a 

 whole may be expected. Viscous phenomena in their entirety, including 

 hysteresis, have a close analogy to the condensation of a vapor. To condense 

 the instabilities to molecular aggregates of small volume takes more pressure 

 than is necessary to release them; i. e., to evaporate them, as it were. If 

 one considers the viscous deformation (cat. par.) as decreasing at a very 

 rapid rate through infinite time, the hysteresis may be considered as an 

 integral part of the viscous phenomenon. The viscous after-effect may 

 then be explained as due to condensation of configurations made unstable 

 or evoked by the heat motion within the body, whereas all the instabilities 

 present at a given time in relation to the applied stress are swept away in 

 the hysteresis phenomenon. 



A great many experiments were now made to endeavor to locate the seat 

 of the yielding within the apparatus. Thus the bifilar was variously attached 

 independently of the weighting appurtenances, the base was clamped and 

 screwed down in different ways, a new rigid bar FF was constructed, etc.; 

 but all these attempts failed to eliminate the discrepancy. Pull on the frame- 

 work and distribution of weights upon it produced no displacement of fringes. 

 Only when weights are placed on the scale pan does yielding (real or apparent) 

 occur; so that it must in some way be connected with the offsets. Replacing 

 the conical ends by sharp darning-needle points was not advantageous. 



Somewhat improved results appeared when the bifilar threads were re- 

 placed by brass strips about a foot long, a half inch broad, and one-sixteenth 

 inch thick, care being taken to insert them without stress. An example 

 of a cycle (e referring to the ocular micrometer) with steel rods may be given 

 5.03 cm., 2r = o-37 cm.). 



P= 5 4 3212 3 4 

 io s Ae/AP = 85 75 60 38 o 25 50 70 85 cm. 

 io 5 AAf/AP = 170 150 120 76 o 50 100 140 170 cm. 



The results of many similar experiments all consistently indicated the same 

 order of yield within the apparatus as before. Data would have been much 

 smoother but for the air-currents in a steam-heated room, which caused the 

 fringes to vibrate. The yielding is usually excessive at the lower loads. 



A final test was made by replacing the slit of the collimator by a glass 

 scale (Chapter I, 5). This is distinctly seen in the telescope and supplies 



