THE AID OF THE ACHROMATIC FRINGES. 



23 



adequate seat for this brittle and slender glass red. There was also evidence of 

 apparent viscosity, to be attributed to the relaxing clutch of the fusible metal. 

 After grinding the outer face of the rod again, experiments were made first 

 between two and four turns and thereafter between two and five turns of the 

 screw. The results are given in figure 26. One notices that the successive 

 cycles gradually close up. After the sixth series there is practical coincidence 

 and the rates per turn of the screw are Ae = 8.g (ingoing) and Ae = 9.8 (out- 

 going). The fringes showed Ae/AAfXio 3 = 22.2. Hence EX io" n = 2g.g/^.e = 

 3.4 and 3.0. This rod showed marked (apparently) viscous contraction, quite 

 apart from the hysteresis of the cycles. Since A#= 10-3/2 2. 2 = io~ 6 X45 cm -> 

 and Ax = o.78AA r , Ax being the contraction of the rod, it follows that A# = 

 io~ 6 X35A<? cm. The rod left for 45 minutes under the compression of five 

 turns of the screw showed an apparent contraction of Ae = 4 scale-parts or io~ 6 

 Xi4o cm., or icT 6 X74 cm. per centimeter of length. More systematically 

 in the following tests (reduced to a centimeter of length) . 



Time 50 80 120 195 minutes 



io 6 Ax o 140 350 787 cm. 



On taking the rod out, however, it became clear that the real reason for this 

 was the gradual yielding of the fusible metal clutch. It is thus probable that 

 the whole length of the glass rod, instead of the part projecting from the 

 fusible metal plug (fig. 7, c), should have been inserted into the equation, 

 making E larger than above given. Hence I returned finally to the sheath 

 method (fig. 7, a) using a thin glass rod, , = 2.54 cm. long, 0.185 cm. in 

 diameter, A =0.0269 cm. 2 . The results are given in figure 27. The graphs 



are nearly coincident, but curved. The mean rates for the higher loads are 

 (per turn of the screw) A0 = i o . 4 (incoming) and Ae = 8 . 6 (outgoing) . Ae/A N X 

 io 3 =28.8 being the fringe factor, EXio~ n = 57.86/^0 = 5.5 and 6.8, respec- 

 tively. Hence here, also, as in the case of the brass rod above, the normal 

 value of the modulus has been reached; i.e., one may expect the data for E 

 to be correct in their absolute values, if the ratio of length of rod to diameter 

 is of the order of i o to i . 



Another series with a different adjustment of the same rod is shown in 



figure 28, the fringes being larger, A0/AA 7 "Xio 3 = 

 lowing mean results: 



Turns of screw 



A0/ 



Triplets gave the fol- 



