30 
Research Bulletin No. 2 
more spikes, each joint of the rachis bearing two spikelets, or as 
meristic variations or repetitions of a rachis bearing two spike- 
lets. The rows are paired, and no case is known where a mem- 
ber of a pair has been aborted. De Vries has regarded the 
number of rows on the maize ear as continuous variations which 
are inherited in a somewhat temporary manner. He came to 
this conclusion from a. selection experiment in which he was able 
to increase the number of rows in the variety with which he was 
working, from an average of about 13 rows per ear to an average 
of about 20 rows per ear in seven years, only to find that the 
selected variety gradually reverted to the type of the original 
variety when selections ceased. We do not question these facts 
but the interpretation that de Tries gave to them. The work of 
East, Emerson, Hayes, and Shull has shown conclusively that 
the reversion in de Vries' selected strains came about simply 
thru the intercrossing of the progeny of plants that were somatic- 
ally alike but genotypically different. The evidence presented 
here shows that the character complex of number of rows per ear 
is made up of a series of cumulative unit factors independent in 
their inheritance. 
Altho there seems to be no good reason why this character 
complex should not be considered to be of the same nature as 
that of other size characters in which variation appears to be 
continuous, it possesses the advantage of being divisible into a 
discontinuous series of numbers. From such a numerical series 
one can obtain some idea of the prevalence and magnitude of 
non-inherited fluctuations. A large number of counts of the 
number of rows of both ears on stalks that bore two ears has 
shown a variation which is usually ±2 rows, altho it may be as 
high as ±4 rows. An ear which appears to belong to any par 
ticular class, therefore, may be pushed by external conditions 
into the wrong class, but the proper class will be one of five in 
which the apparent class is the central one. All ears, however 
do not have the same ability to respond to external stimuli. The 
8-rowed condition, which is the last subtraction form in which 
maize appears, may be selected until the strain breeds very true. 
In a count of the population of an isolated maize field where 
Longfellow, an 8-rowed flint, had been grown for many years. 
4 four-rowed, 993 eight-rowed, 2 ten-rowed, and 1 twelve-rowed 
ears were found. On the other hand, a Learning dent, selected for 
many years and inbred by hand for six generations, still gave a 
range of variation from 12 to 20 rows. Part of this difference in 
behavior shown by races with a low number of rows and races 
with a high number of rows is undoubtedly due to correlation 
with other characters — both physiological and gametic — but at 
the same time it is more reasonable to suppose that an ear 
