626 ADOLF F. VOIGT 



with the integrating type of radiation meters, such as ionization cham- 

 bers and electroscopes. Since these instruments are less sensitive, 

 the number of disintegrations required to give observable measure- 

 ment is large enough so that the statistical error is less than the 

 instrumental errors. With the Lauritsen electroscope, for example, 

 it may be necessary to take the average of five to ten readings in order 

 to obtain an average that seems satisfactorily to represent the mean 

 of the individual values. These values frequently fluctuate consid- 

 erably, which is probably, but not necessarily, due to statistics. If 

 one attempts to measure the activity of an a emitter by placing a 

 source of the proper activity directly inside a Lauritsen one finds 

 very large fluctuations, which are undoubtedly statistical. They are 

 larger than those from /3 emitters because the ionization produced 

 by each a particle in the chamber will be very much greater (ten to 

 fifty times) than that from a /3 ray so that many fewer a rays will 

 give a similar rate of fall of the fiber. 



As the intensity of radiation incident upon it increases each type of 

 instrument reaches a point at which it no longer responds linearl}^ but 

 gives a reading that is too low. Discussion of the limits of the 

 instruments and the necessary correction is given in Section D2 

 for the Lauritsen electroscope and in Section D3 for GM tubes. 

 For other instruments calibration measurements should be made. 



2. Errors in Sample Mounting 



Other possible sources of error are in the lack of reproducibility 

 in sample mounting, position, distribution, and thickness. The first 

 three of these can generally be eliminated by standardizing the 

 mounting procedure and geometrical arrangement. One point to be 

 examined is the sample mount; samples to be compared should all 

 be backed by the same thickness of the same material, unless there is 

 actual experimental evidence to show that, for the particular case 

 concerned, the backing is unimportant or can be corrected for. The 

 necessity for this precaution is the phenomenon of back-scattering. 

 j8 radiation, originally traveling in the direction away from the counter 

 is reflected by collisions with electrons and nuclei in the mount. 

 Such reflections send it back into the counter and increase the count- 

 ing rate over its value on a thin mount or one that back-scatters 

 less. The back-scattering of electrons (/3 rays) is greater as the atomic 

 number of the elements in the mount increases. Up to a point it also 



