In book six of his De Genesi ad Litteram (Literal Commentary on Genesis) St. Augustine addresses the question of whether human souls might have been created simultaneously with the rest of the world at the beginning of creation, with their bodies being formed only later on. Augustine gives two arguments against this view:
first, because of that completion of God’s works, I do not see how these could be understood to be complete if anything was not there established in its causes which would later on be realized visibly; secondly, because the difference of sex between male and female can only be verified in bodies. (De Genesis 6.7.12)
According to Augustine’s second argument, sexual differences are not psyche-logical differences, but physio-logical differences.
Compare this now with Tolkien’s account of the Valar in the Ainulindale:
Therefore the Valar may walk, if they will, unclad, and then even the Eldar cannot clearly perceive them, though they be present. But when they desire to clothe themselves the Valar take upon them forms some as of male and some as of female; for that difference of temper they had even from their beginning, and it is but bodied forth in the choice of each, not made by the choice, even as with us male and female may be shown by the raiment but is not made thereby.
For the Valar, “sexual” differences are more than–because prior–to bodily differences, being a mater of “difference of temper” that is then “bodied forth” afterward in the physical appearance the individual Valar choose for themselves.
The Flame Imperishable 
Tolkien’s account of the sexes of the Ainur is reminiscent of this passage of C.S. Lewis’ Perelandra ch. 16, where Ransom tries to describe the gods Malacandra and Perelandra:
Both the bodies were naked, and both were free from any sexual characteristics, either primary or secondary. That, one would have expected. But whence came this curious difference between them? He found that he could point to no single feature wherein the difference resided, yet it was impossible to ignore. One could try – Ransom has tried a hundred times to put it into words. He has said that Malacandra was like rhythm and Perelandra like melody. He has said that Malacandra affected him like a quantitative, Perelandra like an accentual, metre. He thinks that the first held in his hand something like a spear, but the hands of the other were open, with the palms towards him. But I don’t know that any of these attempts has helped me much.
At all events what Ransom saw at that moment was the real meaning of gender. Everyone must sometimes have wondered why in nearly all tongues certain inanimate objects are masculine and others feminine. What is masculine about a mountain or feminine about certain trees? Ransom has cured me of believing that this is a purely morphological phenomenon, depending on the form of the word. Still less is gender an imaginative extension of sex. Our ancestors did not make mountains masculine because they projected male characteristics into them. The real process is the reverse. Gender is a reality, and a more fundamental reality than sex. Sex is, in fact, merely the adaptation to organic life of a fundamental polarity which divides all created beings. Female sex is simply one of the things that have feminine gender; there are many others, and Masculine and Feminine meet us on planes of reality where male and female would be simply meaningless. Masculine is not attenuated male, nor feminine attenuated female. On the contrary, the male and female of organic creatures are rather faint and blurred reflections of masculine and feminine. Their reproductive functions, their differences in strength and size, partly exhibit, but partly also confuse and misrepresent, the real polarity.
All this Ransom saw, as it were, with his own eyes. The two white creatures were sexless. But he of Malacandra was masculine (not male); she of Perelandra was feminine (not female).
Tolkien’s first version of the Ainulindale from the mid-30s does not mention the “difference of temper”; its revised version from 1948 or before does (HoME V and X respectively). Since Perelandra was published in 1943, it is tempting to suggest that JRRT and CSL may have discussed the subject at some point, as it was obviously one of common interest to them.
Anyway, what I came in to say is that I’ve just discovered this blog, that it’s fascinating and that I intend to keep reading it regularly. Thanks!
Thanks for visiting and for the comment. Yeah, the Perelandra passage is very apropos, and your argument from the chronology of the drafts is indeed very suggestive. Great observation.