282 Ward. — On Relations between Host and Parasite 
normally in the tips of the young leaves just emerging from 
the first leaf-sheath, although the seedlings were of different 
heights at the moment of inoculation. 
Further experience has shown that these assumptions, care- 
fully checked, do not vitiate the conclusions drawn from the 
experiments. 
There was, nevertheless, a possible source of error in the 
large number of plants employed in these experiments. Each 
pot contained from 25 to 100 seedlings, according to the 
size of the ‘ seeds ’ sown and their germinating power. I in- 
oculated about ten to twenty seedlings in each case, and 
judged of the results as merely positive or negative according 
to the subsequent discovery of pustules. I now know that 
it may be of some importance whether the plants are well 
or ill-nourished, and it is quite possible that in some of the 
more crowded pots the results were less convincing than in 
later experiments whence this source of error was eliminated. 
Nevertheless such plants show normal infection. 
The question as to the height of the ‘ spear * above ground 
need not concern us, since subsequent experiments, where 
the heights were duly measured and recorded, showed that 
a given species will be infected readily when only 10-20 mm. 
high, whereas the infection fails on another 50-100 mm. high. 
The results are given in the following Table XII. 
These results point to the conclusion that while the Uredo 
grown on B. mollis can quickly and readily infect B. mollis , 
B. velutinus , B. arvensis , B. secalinus , and B. racemosns — all, 
be it remarked, closely allied forms — it is unable to infect 
B. sterilis , B. madritensis , B. maximus , B. erectns , B. commit- 
tatns , or B. interruptus , or does so with difficulty. 
On the other hand, while the Uredo grown on B. sterilis 
can readily infect B. sterilis and B. madritensis — also closely 
allied — -it appears to fail on the other Bromes, except possibly 
on B. maximus. 
In all cases the control series gave negative results. 
As the Table XIII (p. 284) shows, substantially similar 
results were obtained on repeating this experiment on seed- 
