C. Obscrz ations.
It is beyond the scope of this paper to describe in detail the formation of the sporangium, with its annulus and tapetum, or to describe the degeneration of the tapetal cells. Nor is it neces sary to give aminute account of the mitoses leading up to the formation of the sixteen sporocytes. They all agree with the division of the " archesporium " and the number of the chromo somes apparently remains the same, although they are so numer ous that it is impossible to give the absolute number. From care ful counting in several cases, I estimate the number to be between one hundred and twenty and one hundred and thirty.
The unicellular archesporium, which is destined to give rise to all the spores of the sporangium, can be readily distinguished from the surrounding somatic cells. It is larger, and the chroma tin of the nucleus stains much more deeply than that of the other cells. The chromosomes are large and distinctly looped, and in the metaphase of karyokinesis they are split by a longitudinal division (Fig. 17). After each division the daughter-nuclei pass into the resting stage, during which the cell-walls are com pletely formed and each daughter-cell becomes completely sepa rated. The resting stages are comparatively long and the division stages short.
i. The After the sixteen primary sporocytes are formed the nuclei pass as usual into the resting stage (Fig. I). The nuclei are at first comparatively small in diameter), the chromatin-reti culum does not stain intensely, and there are usually from one to three or more nucleoli in each. Meantime the tapetal cells degener ate, giving room for growth of the reproductive elements. This growth must begin very soon for the cells in the resting stage (Fig. i) are not frequently found. When fully grown, the nuclei measure about 14.5 a in diameter, an increase of nearly 50 per cent. During this enlargement the chromatin reticulum is con verted into a delicate moniliform spireme. This is a single thread of chromatin, very much coiled and interwoven and at first distri buted evenly throughout the nucleus (Fig. 2). From this condi tion of extreme delicacy and expansion the chromatin soon passes into a stage of greater localization and the spireme becomes thicker. Evidence of the beginning of concentration can be seen in Fig. 2 an early stage, where the nucleolus has not disappeared. In a later stage, to which Farmer has given the name " Synapsis," the meshes are drawn towards one side of the nucleus into a much more compact chromatin mass (Fig. 3). This mass next becomes loosened and the filaments more or less isolated. In exceptionally favorable preparations the spireme in this stage is seen to be double (Fig. 3 x).
The concentration of the chromatin at the same time with the thickening of the spireme seems to indicate a coalescence and union of the formerly distinct granules of chromatin in the deli cate moniliform spireme.
2. Pct of telrezd formation; pseudo-reduction.
In the case of animals when the spireme thread breaks up into segments destined to form tetrads, the number of these segments is, in general, half the number of chromosomes in the somatic cells. There is a reduction in number of chromatin masses, but
the nucleus still contains all the chromatin it held at first, so that actual reduction has not yet taken place. (1894) has ac cordingly proposed the expressive term " pseudo-reduction " for this preliminary halving of the number of chromosomes.
Ptcris forms no exception to this rule. The double spireme breaks up into short and well defined chromatin segments (Fig. 5 a) each of which gives rise to a tetrad. The number of these seg ments is difficult to determine; in several cases I counted about sixty. This is about half of the number in somatic cells where, as nearly as I can make out, there are between one hundred and twenty and one hundred and thirty chromosomes. ' It is an inter esting fact that the process of tetrad formation is subject to some variation and does not, apparently, conform exclusively to any one type. This conclusion is based upon the following facts. The spireme segments are, from the beginning, invariably double (Fig. 5 a). The same nuclei contain various modifications of the double segment. Some of them are split in the center while the ends remain connected, giving rise to ring forms (Figs. 4, 5, 19 c). In some there is no separation at all, in others the ends separate, giving rise to " cross " forms (Fig. 6 1 and Fig. 19 a) and in still others one half the segment may slide along on the other half till the ends are no longer contiguous (Fig. 6 d and e). There may be still further modifications of the double segment in the same nucleus (Fig. 5 x). In none of the nuclei which I have examined does any of these types predominate ; and from their various and diverse shapes it is impossible to regard them as developmental stages of a single type. I am forced, therefore, to the conclusion that, in these ferns, tetrads may be formed in a variety of ways. The various methods can be grouped into three types, which I will describe separately as (a) the " ring type ;" (b) the " rod type ;" and (c) the " cross type." The ring type." Almost every primary sporocyte contains from one to several (8 or 9) ring forms in different stages. In some cases the chromatin portion is thin and the opening com paratively large (Fig. 4 c, 5 c, 19 c). In no case is the ring thin and delicate as in Heterocope robusta (Riickert, 1893, Fig. 23). The ring stage begins with a lateral bulging of the two halves of the spireme segment (Fig. 6 i) ; this is followed by the appearance of a furrow at the center. This furrow enlarges until it forms a cir cular space, and, the ends of the segment remaining attached, the chromatin forms a closed ring (Fig. 19 c). The chromatin then begins to accumulate in four parts, each half of the originally double spireme forming two (Figs. 6 J, 19 c and 20 c). These parts become more and more distinct and individualized ; more com pact and tightly packed together, until finally the tetrad is com pleted (Fig. 7). The tetrad is, therefore, derived first, by a lon gitudinal splitting of the spireme segment, and second, by the transverse splitting of the two halves.