[Vor. 3 
68 ANNALS OF THE MISSOURI BOTANICAL GARDEN 
temperatures were taken daily. The results are not conclu- 
sive, as the temperature in the cooler house rose to 21°C. 
on the sixth day and remained high for two days, nor did it 
go back to below the temperature at which infection took 
place. Nevertheless, the symptoms appeared in the warmer 
house March 4, three days before the plants in the eool house 
showed any sign of the trouble, and the cooler soil retarded 
the advance of the fungus proportionately, as may be seen 
from the curves (fig. 20). 
In further experiments on this temperature relation a cold- 
frame was used as a means of maintaining cooler conditions, 
for the greenhouses were all too warm, due to the increased 
intensity of sunlight, especially at midday. The north side of 
the potting-house, where the sun was excluded, made it pos- 
sible to carry these cultures still further into the spring, and 
in all cases the results were the same. 
In the experiment in which the cold-frame was used, a soil 
thermograph of the type manufactured by Julien P. Friez was 
installed. The bulb was imbedded four inches in the soil, and 
the temperature of the soil and air were recorded through- 
out the experiment. The plants were started on March 25 
in six flats of uniformly infected soil, three of which were 
placed in the greenhouse at 25° C. and three in the eold-frame. 
Two pots of greenhouse soil, planted to cabbage, were used as 
controls in each case. The disease appeared first in the green- 
house on April 4, ten days after planting. Seedlings were 
plated from the diseased flats and from the flats in the cold- 
frame on April 9, and in all eases the diseased seedlings 
showed the fungus growing from the stem, while the controls 
remained sterile. On April 13, photographs were made of 
two of the flats—one from the cold-frame showing the healthy 
condition of the seedlings, and one from the greenhouse show- 
ing the ravages due to the attack of the fungus (pl. 2, fig. 
7). The temperature records show that there was an in- 
crease in temperature with the advance of the season, and it 
was due to this increase in temperature that the attack oc- 
curred. The curves (fig. 19) do not show this fact well, as 
they are a record of the maximum and minimum only and do 
