jtinei6,1923 AzoiobcLcter Flora and Nitrogen Fixing Ability of Soils 913 
Even in the laboratory the possibility of contamination could not be 
entirely eliminated. What influence such contamination may have had 
upon the development of Azotobacter from soils that normally con¬ 
tained no Azotobacter is not known. 
Again, it has been shown (j) that Azotobacter can exist for varying 
lengths of time in soils that will not support them indefinitely. Under 
natural conditions, if Azotobacter exists in the proximity of soils not 
containing them, almost constant inoculation due to wind, rain, animals, 
cultivation, etc., is inevitable. The length of time that such introduced 
organisms may remain in a viable condition apparently depends upon 
the intensity of the unfavorable influences. It is, therefore, highly 
probable that Azotobacter may frequently be isolated from soils in which 
they will not function or even exist for a very long period. 
On the other hand, little is known as to how many Azotobacter are neces¬ 
sary to initiate the development of a visible film in laboratory culture 
media. Unpublished data indicate that appreciable numbers are essential 
to the development of a characteristic film. Also, nothing is known as to 
how rapidly diey may disappear from a sample of soil removed from its 
natural environment. It is possible, therefore, that Azotobacter may be 
present in relatively large numbers in a soil and escape observation by 
the methods employed in these investigations. 
In comparing the reaction with the presence or absence of Azotobacter, 
and in obtaining a mathematical expression for this association or 
correlation (the association coefficient), the soils have been divided into 
two groups; those with a Ph of 6,0 or above and those more acid than 
Ph 6.0. This division point has been chosen more or less arbitrarily, 
though the data here presented indicate that it is not far from absolute. 
The association coefficient was obtained by the use of Yule's (14) asso¬ 
ciation coefficient formula, as mentioned elsewhere (5). Associa¬ 
tion coefficients have been calculated both from the data secured by 
the colorimetric and the electrometric Ph determination. In analyzing 
the data presented here it will perhaps be best to examine those from each 
series of soils separately. 
SERIES I 
The data secured from an examination of the soils of Series I are pre¬ 
sented in Table V, rearranged in part in Table VI, and summarized below. 
These soils together with those of Series II are all local soils and were, 
perhaps, collected and handled with more care than those of Series III 
and IV. Samples i to 19 were lost through error before the electrometric 
hydrogen-ion determinations were run. However, in the following sum¬ 
mary the electrometric Ph of samples No. i, 4, 5, 14, and 15 have been re¬ 
garded as above Ph 6.0 and all other soils below No. 20 as more acid 
than Ph 6.0. These 19 soils could not be considered in obtaining the 
numerical average of colorimetric Ph- Likewise samples 85-90, for which 
no quantitative nitrogen determinations were made, were omitted from 
the calculations for average nitrogen fixed. 
