388 MORPHOLOGY. 



axis, and no leaf is ever attached to an axis in this way, but always in 

 the transverse direction : already from this it follows that these projec- 

 tions cannot be leaves. Finally, if the wing of the axis in the folium 

 decurrcns is assumed to be a real foliar part, the analogy would be 

 inapplicable here, since the direction, if from the margin of a hollow 

 axis to the bottom of its cavity, corresponds to the direction from below 

 upwards : now we really know the so-called decurrent leaves, but leaves 

 running up a stem are unheard of. Thus the assumption of the form- 

 ation of the spermophore from the axis appears to be unexceptionably 

 established for this section. The conditions which occur may be 

 arranged according to the following review : 



a. The terminal bud, therefore the innermost and lowest part of the 

 cavity of the germen, may develope as seed-bud (gemmula basilaris unica 

 in g ermine infero, e. g. in the Composite), or the axis may rise up again 

 in the cavity of the germen, and bear the seed-buds as lateral buds 

 (spermophorum centrale in germine infero, e. g. in the Myrtacece). 



b. The internal surface of the cavity of the germen bears the seed- 

 buds in as many lines as there are carpels, without further marking of 

 the spermophores (spermophora parietalia). 



c. From the internal surface of the cavity of the germen project as 

 many ridges, in the same arrangement, the free angles of which bear the 

 seed-buds (spermophora parietalia, e. g. Orchidacece). 



d. The projecting ridges become so broad that they meet in the axis 

 of the cavity of the germen, and thus form spurious septa ; then their 

 borders split into two layers, which, somewhat recurved, project into the 

 two contiguous cells, and each bears seed-buds in its free angle (gemmulce 

 in angulo loculorum interno affixes, e. g. Iridacea). 



Obs. 4. Here 1 have nothing to add, but only to direct attention to 

 what has been said regarding the germen. If I was right then, the 

 matter here is self-evident. 



Obs. 5. On the cases here brought forward I will not venture to 

 judge, because I do not know the complete course of development. In 

 an earlier work (yet trammelled by the spirit of the old school in which 

 I learnt) I have helped myself out with analogies and guesses, which I 

 here expressly recall. True observation and investigation of nature 

 have shown me how this road can never lead to safe conclusions, and in 

 most cases does lead to error, since, to use analogy, we must first have 

 higher principles of unity and universal laws ; and these are just what 

 we as yet are destitute of, and can never be gained in the manner in 

 which it has been attempted hitherto. Therefore, I prefer rather to 

 confess my ignorance, where I know myself not to be warranted by a 

 complete knowledge of the development, than to think only of the 

 present, and purchase the distant possibility of being held an acute 

 observer of nature, with the much closer probability of a most wretched 

 blunder. 



Finally, I have a few general remarks to subjoin. It is a frequent 

 phenomenon for the seed-buds to be seated on two- or many-budded linear 

 spermophores, and since in these cases just as many spermophores as 

 carpels, therefore twice as many rows of seed-buds, exist, this condition 

 has much contributed to nourish the prejudice that the seed-buds arise in 

 rows from the margins of the carpels. With the countless instances of 

 a different kind of structure of the spermophore, the correctness of this 

 relation admitted, the matter would not be of great importance. But 

 we have quite a different explanation afforded us of the bi-seriality of the 



