258 WILBUR WILLIS SWINGLE 
center probably along the original line of fusion of the homo- 
logues; 2) along the line of the secondary split after the fashion 
described by Robertson ('14) and Wenrich ('16) for the Orthop- 
tera. Condensation of the chromatin begins at this stage. In 
early stages of ring formation the tetrads stain rather lightly 
and are somewhat (fig. 17) diffuse, but as condensation proceeds 
they readily take up the basic dyes. The tetrad character is 
obvious from a study of the synaptic ends of the homologues. 
The larger rings in middle prophase stages are generally in the 
form of figure 8's, and this character may be maintained up to the 
metaphase. The smaller tetrads early assume the character of 
the rings; other shapes, such as crosses and y's, and in some cells 
a rod, appear. There is usually a single Y and a single cross- 
shaped tetrad in every spermatocyte, though these may appear 
much like small rings. Other shapes that appear are transitory 
stages in ring formation, or else portions of rings viewed from 
various angles. Large crosses, for instance, sometimes appear 
in early prophase, and are generally true rings viewed 'en face,' 
the arms of the cross being the long synaptic ends of the paired 
chromosomes. Such crosses are not comparable to or to be con- 
fused with true cross-shaped tetrads (figs. 19). 
In sections shghtly overstained, the smaller rings appear solid, 
the synaptic ends being represented merely by rounded knobs. 
At times such rings may even appear like dumbbells, and this is 
notably true of the second-year spermatocytes, i.e., those that 
give rise to true spermatozoa at the second ripening of the germ 
cells of the larvae (figs. 120 to 128). Indeed, it is not improbable 
that those investigators of anuran spermatogenesis who have 
described prophase tetrads as solid and of dumbbell shape were 
perhaps dealing with either overstained or imperfectly fixed 
material. 
Spindle-fiber attachment is non-terminal usually, but may occur 
anywhere in spermatocytes of the first-year larvae showing cen- 
trosomal fragmentation. There are no normal spermatocytes in 
the first maturation cycle, so any discussion of spindle-fiber 
attachment is useless (figs. 29, 31, and 32). 
