214 
E. M. EAST 
balance with very unequal arms. It is rare to find students who 
think of these errors and endeavor to correct them, although such 
correction is as necessary in biology as in physics. Let us see how 
our biological problem fits the rules for the treatment of errors in use 
in experimental physics. 
It was desired to record, in such a manner that they would be 
comparable, numerics that represented the phenotypes of series of 
plants of species of Nicotiana in regard to corolla length and spread, 
sufficiently accurately that genetic analysis of the results might be 
made. 
The investigation was initiated by a series of preliminary measure- 
ments designed to show the practical physical limits to the precision 
of the direct measurements. Repeated measurements of the same 
flowers showed that there were residual errors beyond one millimeter 
in the case of length and two millimeters in the case of spread of 
corolla. Measurement to millimeters was adopted, therefore, although 
these measurements were afterwards thrown into larger classes for 
reasons that can be justified biometrically. 
Then came a study of ontogenetic variation in order that the 
factors affecting such variation might be detected. The factors that 
would naturally occur to anyone who had had experience in growing 
plants were time of planting, physical and chemical condition of the 
soil, moisture, age of plant, flowering period, age of flower, position of 
inflorescence on plant and position of flower in the inflorescence. To 
determine the effect of each of these factors, it was necessary of course 
to eliminate the influence of all the others as far as possible. Since 
the cultures to be compared were nearly always planted at the same 
time, and since this variable is somewhat dependent upon others that 
were under consideration, it was neglected. My cultures have also been 
grown in well-drained soil very uniform in its fertility, but it was 
thought wise to determine how much effect extreme soil conditions 
might have. Several species growing outside in soil of good tilth were 
compared with greenhouse pot cultures. Three-inch, four-inch, five- 
inch and six-inch pots were used in various species, but the treatment 
was uniform for each species. The species were N. tahacum (several 
varieties), N. rustica (several varieties), N. longiflora (two varieties), 
N. sylvestris, N. paniculata, N. acuminata, N. forgetiana and N. alata 
grandiflora. Since only from ten to twenty plants could be grown 
in the greenhouse in most cases, statistical constants were not calcu- 
