496 PROFESSOR W. E. AYKTON, MR. T. MATHER AND Mil. V. E. SMITH: 



be correct within about 5 in 1,000,000. This conclusion assumes absolute constancy 

 of the dimensions of the coils in the interval between measurement and erection, or 

 the same relative change in dimensions. The values of the mutual induction of the 

 coils on fixed cylinder No. 1 and on suspended cylinder No. 1, and of the coils on the 

 No. 2 cylinders were independently calculated by two. of us (T. M. and F. E. S.) in 

 July, 1905 (see Section 10), the difference in mutual induction of the two systems, as 

 calculated, being 0'0062 per cent., that of the first system being the greater. When 

 the ampere balance was completed and the equipment and settings made satisfactory 

 (September, 1905), the difference in mutual induction as found experimentally was, 

 and still is (April, 1907), 0'0054 0'0004 per cent., that of the first system being 

 the greater. Particulars of this experimental determination will be found on p. 515. 

 An experimental estimate of the difference in mean diameters of two coils on a 

 suspended cylinder is 3'5/u, + l/i (p. 516), a value in satisfactory agreement with the 

 difference found by direct measurement. 



Insulation of Coils. The insulation of the helices was next proceeded with. For 

 this purpose an X -shaped framework of wood was mounted on a metal axle and fitted 

 inside the marble cylinder under observation ; the axle was supported on bearings, so 

 that the whole could rotate freely. The marble is semi-transparent, and when an 

 electric lamp is placed inside a cylinder the air gaps between the strands are easily 

 inspected. Under these conditions the appearance of the coils was very beautiful, 

 and close inspection with a lens failed to reveal any defects in the winding. The first 

 measurement of the insulation resistance between adjacent helices indicated it to be 

 of the order of 50 ohms, and the filament of an electric lamp glowed brightly when 

 placed in a circuit containing the two helices and the gaps of what appeared to be 

 marble and air. That the marble was not at fault was shown by tests on the 

 unwound portion of the cylinder, and examination of the gaps with a powerful lens 

 failed to reveal any metallic bridging pieces. In their shortest parts the gaps are 

 0*15 millim. long, and on the fixed cylinders there are four gaps, each nearly 

 93 metres wide ; several days were spent in their examination, and on one occasion a 

 silk thread was passed between neighbouring strands ; the insulation resistance still 

 remained less than 100 ohms. It is unnecessary to describe in detail all the subsequent 

 attempts to locate the leaks. The cylinder and coils were washed with a thin shellac 

 varnish, made by dissolving shellac in ether, but there was no improvement ; after- 

 wards they were washed in ether and then absolute alcohol, but without noticeable 

 effect. The cause of the low insulation resistance was apparent, however, for at the 

 bottom of a porcelain dish containing the used alcohol a fine sediment settled which 

 consisted of minute particles of copper. Apparently the copper strands had a very 

 loose, scaly skin, and thousands of these tiny particles of copper were bridging the air 

 gaps and so diminishing the insulation resistance. The washing with alcohol was con- 

 tinued and the strands lightly brushed with a camel-hair brush, a 32-c.p. lamp being 

 lit through the circuit containing the gaps. Eventually two 32-c.p. lamps were placed 



