16 BULLETIN 911, IT. S. DEPARTMENT OF AGRICULTURE. 
the irregular rows is another indicative characteristic. The shape of 
these unbranched F 2 ears and also of many of the F 3 and F 4 inter- 
mediates strongly suggests the ears of a variety from the Canary 
Islands, the tassels of which have been noted as suggesting Eamosa. 
Other ears with 3 to 10 branches, similar to those found in Mr. Mack's 
variety, resemble the branched-ear variation of the Pawnee variety 
(PL XV) . A similar type of branching has been found in the Texas 
Surcropper variety by Mr. W. TV. Ballard and in the California 
Yellow Flint by Mr. C. G. Marshall. However, in these forms of 
branched ears from other than Eamosa stocks none has been found 
that had more than 4 rows of seeds on the branches, while on many of 
the Eamosa segregates the number of rows of spikelets on the basal 
branches is frequently 8 and sometimes as many as 10 (PL XVI). 
The gradation from typical ramose ears to ears without branches 
affords evidence for the reduced-branched theory of the origin of 
the many-rowed cob (2), although in many of the ears the rows are 
extremely irregular and the abrupt taper from base to tip, with 
the consequent continuous reduction in the number of rows, militates 
against the development of the common cylindrical ear. Indeed, a 
study of these ears raises a serious doubt as to the possibility of de- 
veloping our present regular-rowed ears through a reduction of 
lateral branches to pairs of spikelets. In many instances the inter- 
mediate ramose ears clearly show that the branches reduce to a 
single spikelet instead of a pair, and in consequence each alveolus has 
but one seed; while the irregularity of the whole rachis precludes 
the possibility of determining accurately the number of rows, it is 
apparent that such ears could have an odd number of rows. The 
origin of the original ramose variation in the Learning variety is of 
interest in this connection, since one of the most constant char- 
acteristics of the Learning variety is the poorly formed tip, which in 
many ears appears to have been artificially joined to the main portion. 
It is difficult to visualize the common ear pf maize as having arisen 
through the reduction of the branches of the inflorescences as they 
are now constituted, since there is little regularity in their present 
arrangement. If the branches of a normal maize tassel were reduced 
to paired spikelets, these spikelets would be arranged in a very 
irregular manner on the rachis and would bear little resemblance to 
the regular-rowed ears. If the reduction is assumed to have occurred 
at an earlier period when the branches of the inflorescence had a dis- 
tichous arrangement similar to the present arrangement of the vege- 
tative branches, the resulting ear would have but four rows, and the 
problem of their increase to the present number remains. 
Hackel has explained the increase in the number of rows as the result 
of a coalescence of 4-rowed branches, a theory supported by numerous 
bifurcated and bear's-foot ears. Collins (3), in studying teosinte- 
