282 The Increase in the Number of Erythrocytes with Altitude 
This table shows that the mean or average of these twenty observations was 
5,930,000, with a probable error of + 19,000 and a standard deviation of 127,000. 
The probable error of the mean is small compared with the mean, showing 
that the error of technique was also small. The counts were registered only to 
the nearest 100,000. 
Fig. 1. Kesults of 20 Observations on one Individual. 
8 
7 
6_ 
5. 
4. 
3. 
2. 
1. 
0 
57 5-8 5-9 60 6-1 6-2 
Erythrocytes in Millions. 
(b) If a theoretical expression of the probable distribution of erythrocytes 
per unit of volume can be obtained this would form, when compared with the 
actual state of affairs, a valuable test of observational and technical accuracy. 
We have utilized for our purpose a computation given in an article written under 
the name of "Student" in Biometrika (Vol. v. p. 351). The computation is 
expressed in the formula 
1 + ra +_ + ..._ + . 
2 1 rl 
this being a very good approximation to the point binomial which would still more 
closely describe the distribution. 
Table I gives the results of a count of 400 squares of a Thoma-Zeiss 
haemocytometer. The blood in this case was for convenience diluted 500 times. 
TABLE I. 
0 
1 
2 
3 
4 
5 
6 
7 
8 
Erythrocytes per square 
30 
72 
105 
104 
52 
27 
8 
1 
1 
Number of squares, observed 
33 
82 
103 
86 
52 
26 
11 
3 
2 
Number of squares, calculated 
It seems evident from the above table that the observed and calculated figures 
do not differ greatly. And when we apply the necessary statistical tests {vide 
:: »(,-— mean of frequency distribution. 
