THE AID OF THE ACHROMATIC FRINGES. 51 



H refers to amperes. The Siemens dynamometer, with its large resistance 

 (480 ohms) was excluded. The circuit resistance was now about 180 ohms. 



The data of s and number of turns n\ found for an inclosed resistance of 

 R = 1000 ohms and R = o are given in figure 54. They are mean results, as the 

 current fluctuated for incidental reasons difficult to enumerate. This was 

 particularly the case for coil B. When newly inserted (R = o) the band- width 

 was apt to be above 5=55; but after long vibration it fell off to 5 = 40 or even 

 less. Fresh washing of the interrupter made no difference. 



As this discrepancy disappears in the lapse of time, if the coil is not used, 

 and as it is present markedly only in connection with the larger coils, it is 

 probably a temperature effect resulting from heating the coils of the bobbin 

 of the telephone. 



Figure 54 shows that for R= 1,000 ohms additional resistance, the vibrator 

 indicates 0.016 scale-part per turn of wire in the secondary ; for R = o the datum 

 is 0.063 scale-part per turn. In each case, however, the graph which is fairly 

 linear, begins with an initial 5 of 3 and 5 scale-parts, respectively. This 

 recalls the similar experiences above. It can not be initial band- width, as 

 this did not exceed i scale- part. 



34. Heavier armature and less damping. With the observer still within 

 easy reach of the apparatus, a further increase of sensitiveness may be gained 

 by removing the iron plates gg' of the lens carriage (figs. 41, 42) and soldering 

 to it on the side of the telephone i, a heavier piece of soft iron (about a square 

 centimeter in area and 0.2 cm. thick). The maximum band-width was 

 reached about as soon as before and the instrument was not much more sub- 

 ject to tremors during manipulation. It was curious to observe that when the 

 current is closed any existent band-width sometimes expands further, while 

 at other times it first closes and then expands, depending on the phase in which 

 it happens to be caught. Having been adjusted for resonance at the beginning 

 of the work, the wires were thereafter left without further interference, assum- 

 ing that the same resonance adjustment belongs to each observation. Though 

 this is not strictly true, it greatly facilitates the measurements. With the 

 new apparatus, moreover, the cyclic phenomena of figure 45 were not noticed. 



In this way the sensitiveness was about doubled, as compared with the 

 former values, though in the repetition of the experiments with coils the band- 

 width of about 5 = 30 scale-parts per milliampere was not exceeded. The 

 results of the new tests of the coils 30, 40, W, B, G, 2G, and D, are given in 

 figure 55, the resistances inserted being respectively 1,000 and 2,000 ohms, 

 while the resistance of the circuit itself (in the absence of the Siemens 

 dynamometer) did not exceed 200 ohms. The G coil contained two helices 

 wound side by side, which here could be used in parallel and in series. The 

 curve for 1,000 ohms is fairly linear (without beginning at the origin, however), 

 with the exception of the datum for the D coil, which is small for the reasons 

 given above. The series result 26" exceeds the individual results for G. At 

 2,000 ohms the graph was less satisfactory and, even if the D coil is excluded, 

 could not be considered straight. One reason for this is the fact that but a single 



