THE AID OF THE ACHROMATIC FRINGES. 19 



flat with a flat carborundum stone, parallel to the abutment B, figure 7. It 

 was also kept under eight turns (about 14 kg.) for 2 hours. The data are given 

 in figure 14, series (8) and the mean results of the straight lines are (Ae = 



(Returning) Ae per turn, 3.6 

 EXio- 11 7.2 



4.2 (outgoing) 

 6.2 



The return series has as usual the larger modulus; but the outgoing series 

 is here above the return series in Ae. One should notice that the mean dis- 

 placement Ae per turn is only about Ae = ^.g equivalent to A7V = 23Xio~ 5 cm. 

 orA^=i8Xio- 5 cm. Since the true value or Ae should be 2.6 instead of 3.9, 

 Ae'= 1.3, equivalent to AN = 0.00008 cm. or Ax = 0.000,06 cm. (i.e., about the 

 wave-length of light) is lost in the apparent yield of the apparatus per turn of 

 compressing-scre w . 



The rod was now left under stress (i4kg.) for about an hour and then showed 

 the displacements Ae given in figure 14, No. 9. The slope of the straight part 

 of the curve, i.e., the modulus, has remained about the same. One may note 

 also that moderate loads (below 7 kg. on 0.035 cm 2 .) are best adapted to bring 

 out linear graphs, quite up to the absence of all load. 



On removing the rod from its sheath there was no evidence of any change 

 of shape and it still fitted the sheath loosely; only at the neck of the sheath 

 the rod showed ringlike chafing. 



9. The same, continued. By way of contrast a thick solid brass rod, 

 L = 2.34 cm. and 0.376 cm. in diameter, A = O.III cm 2 ., was now put into the 

 sheath and tested, the aim being to determine the limit of measurement. 



We should thus have (A/AA^Xio 3 = 16 here) 



io-" = m. 7 8-^i- 93Xl6 A = 7 .i8n/Ag 

 o.in nX-707Ae 



The first and second series are given in figure 15, (i) and (2), and the mean 

 data are: 



Ae per turn 2.7 4.9 3.7 4.6 scale-parts. 



EXio" 11 2.6 1.5 1.9 1.5 



These values of E are much too small, as was to be expected. The apparatus 

 yields much more than in the preceding case. Ae per turn should here be 

 but 0.7 scale-parts. 



The foot of the rod r, figure 7, was now ground down with a small flat car- 

 borundum stone placed between r and B and slid up and down under spring 

 pressure. This is the best way to obtain parallel surfaces. Results so found 

 are given in figure 15, (i) to (4), beginning with a decreasing thrust. If 

 we take the mean rate for the higher pressures (the rod loosens below four 

 turns), the data came out (Ae/ANXio 3 = 18) as follows: 



Series (i) (2) (3) (4) 



Ae per turn ... 2.7 1.2 1.7 



Xio- u ... 3.0 6.8 4.5 



