—— — 
122 
A difference of 0.97+.28 between the means of two samples from the 
same specimen (Nos. 52 and 53)—a difference more than three times the 
probable error—shows clearly the difficulties of sampling, and that such 
differences between samples of the same species grown under the same con- 
ditions may be expected. The differences in several instances, notably 
between Nos. 44 and 49, 44 and 48, and 46 and 51, are no greater than those 
between two samples of the same specimen and may well be due to sampling, 
and to this extent show the fungus, over a wide geographic range, to be 
remarkably uniform. In several other instances, however, there is a wide 
difference of means, above the probable error—notably in all cases involv- 
ing sample No. 53. These differences are often four, five, or six times the 
probable error, and occasionally run as high as eleven or twelve times the 
probable error even with this remarkably uniform fungus. While these 
differences may in part be attributed to sampling they probably represent 
also morphological changes due to environmental differences, and differ- 
ences of nutrition or humidity, but do not necessarily indicate racial dif- 
ference in the fungus. 
To determine whether various cereals, autoclaved, influence conidial 
length differently, plates of H. No. 1 were prepared under standard condi- 
tions except that in the same Petri dishes were placed shoots of wheat, rye, 
barley and corn. The resulting graphs of conidial length are given in Fig. 
M. The differences in means are as follows: 
On rye and wheat, 0.40 + .29 
On rye and corn, 0.45 + .20 
On wheat and corn, 0.02 + .22 
The mean length on rye, corn, and barley is in close agreement with 
‘that on wheat, and, apparently, under these conditions the species of 
shoots counts for little in its influence on conidial length. 
Conidial-length graphs (Fig. N) made from H. No. 1 grown on fresh 
wheat-stems, on young wheat shoots, on wheat leaves, and on young wheat 
plants, all autoclaved in test-tubes with a few centimeters of water, show a 
considerable increase over those under standard conditions (Graph 42, 
Fig. K); also, in Graphs 58, 60, and 61 (Fig. N), they show a much larger 
standard deviation and coefficient of variability, probably due to the va- 
riable humidity under these conditions. The small number of conidia 
measured, and the lack of control over humidity may be presumed to ac- 
count for such variation as is seen. 
Live wheat inoculated in rag doll showed at the 6th day 100% infec- 
tion. These infected seedlings were placed in a Petri dish on moist filter- 
