16 BACTERIA IN RELATION TO PLANT DISEASES. 
assumed cause, the different number of cases in the two groups of plants being 
accidental variations. If, in such a locality, only a very few plants are inoculated 
and a few held as checks, the evidence becomes still weaker and would not be con- 
sidered entirely conclusive even though all of the inoculated plants should contract 
the disease and all of the checks should remain free, since in a region subject 
to a given disease five or six healthy plants may sometimes be found in prox- 
imity to five or six diseased ones, although all may have appeared healthy earlier in 
the season. ‘The case is quite different if out of 100 control-plants and 100 inocu- 
lated plants 95 per cent of the latter and only 2, 5, or 10 per cent of the former 
contract the disease. It then becomes a question of probability which may be 
converted into reasonable certainty by several repetitions of the experiment with 
like results. Of course, the ideal experiment is one in which all the inoculated 
plants contract the disease and none of the control-plants, and in which a large 
number of plants has been used so as to exclude all possibility of the results being 
due to anything but the organism used. 
Whenever the disease occurs naturally in the vicinity selected for the experi- 
ments, too much emphasis can not be laid on the necessity of having numerous 
inoculated plants and numerous controls, and on the desirability of repetitions of the 
experiment in different years and under different local conditions. It is important 
also that the inoculated plants should be under healthful conditions, z. ¢., under 
conditions as nearly natural as possible. For example, proper (natural) conditions 
would be much more nearly attained by inoculating vigorous plants growing in the 
open air or in well-kept greenhouses than by inoculating parts of the same plants 
cut away from the stems and kept under bell-jars. It is conceivable that inocula- 
tions which would succeed very well under the conditions last named, especially 
at abnormally high temperatures, might entirely fail when under a more natural 
environment. . 
Not one of these four requirements can be omitted safely. A chain of evidence 
is not stronger than its weakest link. Particular stress, therefore, ts laid on being 
able to produce at will the characteristic signs and lesions of the disease in healthy 
plants by inoculation with pure cultures of a given sort; also on the re-tsolation of the 
organism from the artificially-infected plants after they have become diseased ; on the 
subsequent proper behavior of the organism in nutrient media ; and on its ability to 
produce the disease when again tnoculated. 'This is the whole thing in a nutshell. 
The experiments must be continued until there is no doubt whatever as to the 
pathogenic or non-pathogenic properties of the organism. ‘Almost certainly path- 
ogenic” always leaves room for grave doubt in the mind of every thoughtful reader. 
As a rule, the re-isolations should be made at a considerable distance from the point 
of inoculation, particularly if there is any doubt whatever as to the identity of the 
physical signs, since saprophytes have been known to live in plant tissues for a 
considerable number of weeks near the place of inoculation, and, if abundant, 
might cause various disturbances of nutrition without being the pathogenic organism 
sought for. For example, one would be more likely to obtain the cause of the 
disease in pure culture by attempting isolations from a plant in the stage shown in 
— 
_—— 
