XVII. RADIOACTIVE TRACERS 623 



Commercial tubes with window thicknesses down to 1 .5 mg. per square 

 centimeter are available (84,85, etc.) at somewhat higher prices than 

 are charged for those willi thicker windows. The range of the 

 C^^/3rays is 25 mg. per square centimeter and following a rough rule 

 that range is about eight times the half-thickness of jS radiation at 

 the beginning of the absorption curve, ~ 3 mg. per square centimeter 

 of total absorber would cut the intensity to half, mg. per square 

 centimeter to one-fourth, and so on. An estimation of the loss to 

 be expected can be made on this basis. 



A rather expensive screen wall counter based on a design by 

 Libby (33) is commercially available (85). Samples are introduced 

 into the vacuum chamber and the cathode is a cylindrical screen be- 

 tween the sample and the center wire. Since only one sample can 

 be measured at a filling, its use is quite cumbersome. The design 

 of counters that can measure the tracers while they are in solution 

 has been undertaken by a number of workers and is common practice 

 in some laboratories. The case for dipping counters is discussed by 

 Bale (38) and by Solomon and Estes (39). 



E. ERRORS AND CORRECTIONS IN MEASUREMExNTS OF 



RADIOACTIVITY 



Interpretation of the data from measurements of radioactivity 

 requires a knowledge of the possible errors, the corrections that need 

 to be made, and the limits of accuracy of which the method is ca- 

 pable (40) . Since the corrections need not be made to a degree greater 

 than the inherent accuracy, a discussion of this point will be given 

 first. 



1. Statistics of Counting 



It is important in the interpretation of the data to be able to decide 

 if a result is statistically significant. Although the subject of statis- 

 tics is one of considerable complexity and is beyond the range of this 

 chapter, a brief discussion of this point follows (4-1) ■ 



The disintegration of a large number of the same kind of radio- 

 active atoms is randomly spaced in time. The counts observed from 

 a source containing such atoms will have a random distribution obey- 

 ing the Poisson distribution law. To avoid some of the difficulties 

 in applying Poisson distribution, the normal or Gaussian distribution 



