THE AID OF THE ACHROMATIC FRINGES. 67 



47. Narrow bifilar. After obtaining the favorable results just described, 

 it seemed obvious that the sensitiveness could be further increased by dimin- 

 ishing the distance between the bifilar wires. Accordingly, with the same 

 inductor (fig. 71), the above wires (diameter 0.022 cm.) were adjusted at but 

 1.5 cm. apart, by decreasing the diameter of the lower pulley. A few other 

 modifications were added. The results, however, were disappointing through- 

 out. In view of the relatively greater inertia of the system, the resonance 

 deflection (band-width in the telescope) was therefore difficult to find. Any 

 incidental quiver of the vibrator decayed very slowly. It was necessary, 

 moreover, to use relatively slow vibrations (n = 10 and less) . The wires broke 

 before the higher frequencies were reached. The maximum sensitiveness 

 obtained was about 32 scale-parts for coil No. 10 and ^ = 300 ohms, so that 



300X32 



_ 



go to a scale-part of band-width. This is less than the above result. 



The preceding vibrator (umbrella-rib) was now replaced by a thin iron 

 wire about i mm. in diameter. This with coil No. 10 and n= 10 to 15, gave 

 but 23 scale-parts for the maximum band-width, so that this result is worse 

 than the preceding. A glass vibrator made of a thin-drawn light tube, with 

 iron plates cemented to the end, was next tested. The maximum in this case 

 was reached much more rapidly, but in a great variety of tests and modifica- 

 tions it did not exceed 10 scale-parts. Fast and slower vibrations were tried, 

 without much difference. This vibrator would have been steady enough for 

 use on the interferometer; but because of the lack of sensitiveness, trial was 

 not worth while. 



I next inserted a thicker bifilar wire (diameter 0.036 cm.). The results 

 were not much better. In fact, with the iron-wire vibrator I obtained a maxi- 

 mum band-width of 23 scale-parts, the same as in the preceding case. The 

 vibrator was much steadier. As there seemed to be no promise of success, 

 further experiments were here also abandoned. 



A few incidental tests were made. Thus with the telephones in series, the 

 band-width of 23 scale-parts fell off to about one scale-part when the telephones 

 were joined oppositionally. This is a good test on the symmetry of adjust- 

 ment, as the rotation of vibrator is changed to translation, broadsides onward. 



Again, the auxiliary audible telephones responded with about equal loud- 

 ness when the telephone circuit was closed with 250 ohms and when a con- 

 denser of i microfarad capacity was inserted. But the vibrator reacted in the 

 former case (resistance and self-induction) with a deflection of 23 scale-parts, 

 whereas in the latter (capacity) the response was at most 2 scale-parts; and 

 this required a slightly different tension of wire. The metallically closed 

 circuit therefore affected the vibrator at least 12 times more strongly than 

 the oscillation through the capacity of i microfarad. Small capacities like o.i 

 microfarad fail to influence the vibrator, though to the ear the sound is quite 

 loud. The capacity would have to be increased to 10 or 20 microfarads for 



