This applies also to Letter 629, written after Darwin had learned Fritz Muller's theory, which is obscured by some errors or slips of the pen.)
LETTER 620. TO G. BENTHAM. Down, February 3rd [1862?]
As you so kindly helped me before on dimorphism, will you forgive me begging for a little further information, if in your power to give it? The case is that of the Melastomads with eight stamens, on which I have been experimenting. I am perplexed by opposed statements: Lindley says the stamens which face the petals are sterile; Wallich says in Oxyspora paniculata that the stamens which face the sepals are destitute of pollen; I find plenty of apparently good pollen in both sets of stamens in Heterocentron [Heeria], Monochoetum, and Centradenia. Can you throw any light on this? But there is another point on which I am more anxious for information. Please look at the enclosed miserable diagram. I find that the pollen of the yellow petal-facing stamens produce more than twice as much seed as the pollen of the purple sepal-facing stamens. This is exactly opposed to Lindley's statement--viz., that the petal-facing stamens are sterile. But I cannot at present believe that the case has any relation to abortion; it is hardly possible to believe that the longer and very curious stamens, which face the sepals in this Heterocentron, are tending to be rudimentary, though their pollen applied to their own flowers produces so much less seed. It is conformable with what we see in Primula that the [purple] sepal-facing anthers, which in the plant seen by me stood quite close on each side of the stigma, should have been rendered less fitted to fertilise the stigma than the stamens on the opposite side of the flower. Hence the suspicion has crossed me that if many plants of the Heterocentron roseum were examined, half would be found with the pistil nearly upright, instead of being rectangularly bent down, as shown in the diagram (620/4. According to Willis, "Flowering Plants and Ferns," 1897, Volume II., page 252, the style in Monochoetum, "at first bent downwards, moves slowly up till horizontal."); or, if the position of pistil is fixed, that in half the plants the petal-facing stamens would bend down, and in the other half of the plants the sepal-facing stamens would bend down as in the diagram. I suspect the former case, as in Centradenia I find the pistil nearly straight. Can you tell me? (620/5. No reply by Mr. Bentham to this or the following queries has been found.) Can the name Heterocentron have any reference to such diversity? Would it be asking too great a favour to ask you to look at dried specimens of Heterocentron roseum (which would be best), or of Monochoetum, or any eight-stamened Melastomad, of which you have specimens from several localities (as this would ensure specimens having been taken from distinct plants), and observe whether the pistil bends differently or stamens differently in different plants? You will at once see that, if such were the fact, it would be a new form of dimorphism, and would open up a large field of inquiry with respect to the potency of the pollen in all plants which have two sets of stamens--viz., longer and shorter. Can you forgive me for troubling you at such unreasonable length? But it is such waste of time to experiment without some guiding light. I do not know whether you have attended particularly to Melastoma; if you have not, perhaps Hooker or Oliver may have done so. I should be very grateful for any information, as it will guide future experiments.
P.S.--Do you happen to know, when there are only four stamens, whether it is the petal or sepal-facers which are preserved? and whether in the four- stamened forms the pistil is rectangularly bent or is straight?
LETTER 621. TO ASA GRAY. Down, February 16th [1862?].
I have been trying a few experiments on Melastomads; and they seem to indicate that the pollen of the two curious sets of anthers (i.e. the petal-facers and the sepal-facers) have very different powers; and it does not seem that the difference is connected with any tendency to abortion in the one set.