SIGNIFICANT ACCURACY IN RECORDING GENETIC DATA 21 5 
lated, for I have not the faith of Goodspeed and Clausen in probable 
errors based on nine or ten observations (see their tables II a, b and 
III a, b). Averages of five flowers per plant taken when first in full 
flower, however, indicated means within a millimeter of each other 
for length and within two millimeters of each other for spread of 
corolla for over half of the species, when compared with the sister 
plants in the field. The greatest difference was in a N. alata grandi- 
flora test where the starved plants showed an average of about 5 mm. 
shorter and 7 mm. narrower flowers. Hybrids were also tested. 
As I do not consider it necessary to cite figures endlessly where they 
serve so little purpose, however, only a table of results on a cross 
between two varieties of N. longiflora is given, the field records and 
the pot records being made by different observers. The general 
Table I 
Frequency Distribution for Length of Corolla in Cross between N. longiflora Varieties 
Designation 
37 40 43' 46 I 49 52 55 
No. 383 field. . 
No. 383 pots. . 
No. 330 field . . 
No. 330 pot. . . 
(383 X 330) 
F3A field.... 
Ditto, pot 
(383 X 330) 
F3B field.... 
Ditto, pot 
(383 X 330) 
F3C field.... 
Ditto, pot 
(383 X 330) 
F3D field . . . 
Ditto, pot 
(383 X 330) 
F3E field . . . . 
Ditto, pot 
Class Centers in Millimeters 
15 
64 67 70 73 76 7Q 82 
70 19 10 
3 • 
2 8 
33 
41 
19 
32 
83 9, 
effect of starvation can be seen even without having the means calcu- 
lated. A comparatively small number of observations were made 
on each population, but they serve as samples of the frequencies found. 
Certainly no mafked decrease in size is apparent, and since the vegeta- 
tive organs of the pot-grown plants varied from one half to one fifth 
the size of those in the field (linear dimensions), it seems that one 
