July 8, 1897] 



NATURE 



235 



§ 7. Similar experiments were made with higher voltages 

 measured by the vertical electrostatic voltmeter, and we found 

 that when the flame was three or four centimetres above the 

 point, there was very rapid discharge ; but when the flame was 

 60 centimetres or more above the point, the leakage from 3500 

 volts was practically the same as if the flame was not lit. 



In place of the metal point, a round disc of zinc, 8 centimetres 

 in diameter, was fixed, as shown in Fig. 3, to the end of 

 another steel wire of the same length ; and leakage from it 

 to the flame above it, observed. For the same distance between 

 the flame and either the point or the metal disc, the rate of 

 leakage through the same difference of potential, was less for the 

 point than for the disc. Thus with the flame 25 centimetres 

 above the point the time of drop from 3CXX) volts to acxx) volts 

 was I min. 53 sees., and with the flame the same distance above 

 the horizontal plane of the disc the time of drop from 3000 

 volts to 2000 volts was I min. 14 sees. This is a very important 

 result. 



§ 8. Experiments were next made to find if, and if so, how 

 much, the leakage is diminished by putting non-conducting plates 

 of glass, paraffin, mica, between the point or disc and the flame. 

 At a corner of each plate was pasted a little square of tinfoil, so 



Fig. 3. 



as to prevent any electrification of the non-conducting substance 

 by handling. The.se pieces of tinfoil were always kept metal- 

 lically connected with the sheath of the electrometer. Each 

 plate was fixed with its under surface i cm. above the steel point. 

 In preliminary experiments (of which a continuation is deferred 

 until the insulation of the electrometer is made practically perfect 

 by coating its vulcanite insulators with paraffin) the following 

 numbers were obtained : — 



I. Glass Plate 18 cms. by 19 cms. by 0*3 cf7i. 



Distance of 



flame above 



point 



Time of fall 



from 3000 to 



2000 volts 



Sec. 

 30 



5 

 7 



Insulation test, with no flame. 

 Flame lit : no intervening plate. 

 ,, ,, glass plate between. 



II. Mica Sheet 18 cms. by 9 cms. by 0'\ cm. 



Insulation test, with no flame. 

 Flame lit : no intervening plate. 

 ,, ,, mica sheet between. 



NO. T445, VOL. 56] 



III. Paraffin Plate 11 cms. hy 11 



rtWo75 ''"'• thick. 



No tlanu-. Insulation test. 

 Flame lit : no intervening plate. 

 ,, ,, paraffin plate between. 



We hope to return to the investigation with the insulation of 

 the electrometer perfected ; and to determine by special experi- 

 ment, how much of the fall of potential in the electrometer in 

 each case is due to the electricity of opposite kind induced on 

 the uppermost surface of the non-conducting plate, and how 

 much, if any, is due to leakage through the air to the metal 

 disc or point below. 



§ 9. To test the quality of the electrification of both sides of 

 the non-conducting plates of glass and paraffin, a thin copper 

 sheet was fixed to one of the terminals of a quadrant electro- 

 meter, as represented in Fig. 4, where A is the plan of plate 

 c attached to the electrometer, and B is the plate of paraffin or 

 glass under test. 



In the primary experiment (Fig. 3) the non-conducting plate 

 was fixed in a horizontal position one centimetre above the 

 electrified metal (point or disc), and eleven centimetres below 

 the flame. A charge was given to the metal, to raise its 

 potential to about 3500 volts. After some minutes, generally 

 till the potential of the metal fell to 2000 volts, the non-con- 



\ 



ducting plate was removed and placed, as shown in Fig. 4, above 

 the metal plate c attached to the quadrant electrometer, and 

 the deflection was observed. For a thin piece of glass (0*3 cm. 

 thick) the whole effect of the two sides was negative when the 

 electrified metal point or disc had been charged positively and 

 vice versd. But on putting two plates of glass above the 

 electrified metal, we found the top plate to be oppositely charged 

 and the under plate to be charged similarly to the point or 

 disc, but not so highly. We found corresponding results with 

 a plate of paraffin 075 cm. thick, and with two plates of 

 paraffin 0*5 cm. and 075 cm. thick. When a plate of paraffin 

 3 '25 cms. thick was used, we always found the top face charged 

 oppositely to the charge of the metal, whether disc or needle- 

 point, and the under face charged similarly to the metal below. 

 Thus the apparent total charge of the two faces of a thin non- 

 conducting plate is due to the fact that the face of the plate 

 away from the electrified metal is more highly charged oppositely 

 than the face next the metal is charged similarly. 



A NEW LAW OF HEREDITY. 

 HTHE truth of a law of heredity proposed by Mr. Francis 



•*■ Galton, has been verified in particular instances, in a 

 memoir 1 read by him before the Royal Society on June 3. 



He first put forward the law, with hesitation, in his book 

 "Natural Inheritance" (Macmillanand Co., 1889), page 134, 

 because it was founded at that time almost wholly upon h priori 

 grounds. Now, being found to hold good in a large group of 



1 " The average Contribution of each several Ancestor to the total 

 Heritage of the Oflfspring," by Francis Galton, D.C.L., Sc.D., F.R.S. 



