by a few minutes' observation, and to a much higher degree 

 with greater care ; it is, perhaps, the easiest of the measure- 

 ments made in these experiments. By far the greatest diffi- 

 culties which T find in the determination of the stopping 

 power of a gas lie in the purification and analysis of the gas. 

 On the other hand, the abscissa I is much more diffi- 

 cult to measure. It is affected by variation in the sensitive- 

 ness of the electrometer, by leakage through the insulators, 

 by variation of the dimensions of the apparatus, and its true 

 value is not given unless enough electric force is applied. 

 None of these things affects the range. But it is not merely 

 in the details of measurement that these two quantities differ. 

 They appear as physical constants to be in distinct cate- 

 gories ; so far, that is to say, as can be observed at present. 

 The stopping power of an atom is a constant of the 

 atom, unaffected by its association with other atoms in mole- 

 cular structure, independent of pressure and temperature. 

 In a paper by Mr. Kleeman and myself ('Phil. Mag.," Sejot., 

 1905), we gave a list of the stopping powers of various sub- 

 stances, and since then we have made many other experi- 

 ments in the same direction. In no case have we found a 

 departure from the additive law which was not within the 

 errors of experiment. That is to say, the range of the a par- 

 ticle in a given gas can always be predicted from the com- 

 position of the gas molecule. Not only so, but the stop- 

 ping powers of the various atoms are very nearly propor- 

 tional to the square roots of their weights, so that a simple, 

 if approximate, law covers all the phenomena. It even 

 seems justifiable now to go one step further. If the list in 

 the paper quoted be examined, or the more comprehensive 

 list in Table A, it will be found that the stopping powers are 

 systematic even in their slight departures from the square 

 root law. For, whilst dependent mainly on the square roots 

 of the weights they have a leaning towards the weights them- 

 selves. We did not call attention at the time to this fact, 

 for we thought it might be a spurious effect. But it has ap- 

 peared so regularly in all further determinations that it seems 

 right to note it, and to attempt an explanation of its physi- 

 cal meaning. 



If we assume the correctness of the explanation already 

 given of the square root law, viz., that the a particle spends 

 energy for the most part on tearing away electrons from their 

 attachment at the edges of the atom discs, then the natural 

 complement to this is the further assumption that electrons 

 in all parts of the atom disc may be disturbed to vibration 

 by the passage of the a particle, which latter, therefore, 

 spends a small amount of energy in simple proportion to the 

 b2 



