1911] on a New Method of C hemic al Analysis. 141 



When these rays strike ao^ainst a photographic plate, they affect the 

 plate at the point against which they strike, and thus when the plate 

 is developed we have a permanent record of the deflections of the rays. 

 The methods of taking these photographs and the details of the ex- 

 periment are described in my paper in the ' Philosophical Magazine,' 

 Feb. 1!)11. The values of A and B can be determined accurately by 

 the methods I have given in previous papers, and hence if we measure 

 on the photographs the values of ON and PN, we can determine 

 the value of m/e. If we wish to compare the values of m/e for two dif- 

 ferent rays, it is not necessary to determine A and B, all we have to 

 do is to measure the values of ON and P N" ; and thus the photo- 

 graph alone gives us the means of comparing the value of m/e. 



PN- 

 For the same type of carrier mle is constant, so that ^^r=r- is 

 ^^ ' ON 



constant whatever may be the velocity, and therefore the locus of P, 

 i.e. the curve traced on the photographic plate by this carrier, is a 

 parabola. The reason we get a curve instead of a point is that the 

 rays are not all moving with the same velocity, and the slower ones 

 suffer greater deflection than the quicker ones. Each type of carrier 

 produces its own line on the plate, and there are as many curves 

 on the plates as there are kinds of carriers ; from an inspection of 

 the plates we can find not merely the number of kinds of carriers, but 

 from the dimensions of the curves we can at once determine the 

 atomic weight of the carrier, and thus determine its nature. This 

 is one of the great advantages of this method. To illustrate this 

 advantage, let us compare the method with that of spectrum analysis. 

 If the spectroscopist observes a line unknown to him in the spectrum 

 of a discharge tube, the most he can deduce without further investi- 

 gation is that there is some unknown substance present in the tube, 

 and even this would be doubtful, as the new line might be due to 

 some alteration in the conditions of the discharge. But if we observe 

 a new curve in the positive ray spectrum, all we have to do is to 

 measure the curve and then we know the atomic weight of the sub- 

 stance which produced it. To take an example, I have photographed 

 the positive ray spectrum for nitrogen prepared from the atmosphere 

 and that for nitrogen prepared from some nitrogenous compounds, 

 and have found that the former contains a line * which is not in the 

 latter, and that the value of m/e for this line is 40 times that for the 

 atom of hydrogen. We thus know that atmospheric nitrogen con- 

 tains an element of atomic weight 40, which is not present in 

 chemical nitrogen — this element is, of course, argon. We might l)y 

 ordinary spectrum analysis have found lines in the spectrum of 

 atmospheric nitrogen which are not in the spectrum of chemical 



* As a matter of fact, there is a second, very faint line for which m/c is about 

 20 times that for the atom of hydrogen. This is probably due to an atom of 

 argon with two electric charges. 



