A few days ago lc argued that "the ceiling for female sexual attractiveness is not very high."

I have no strong opinion but this interests me and I think it could be investigated in a more systematic manner. 

Operationalised Definition of "Female sexual attractiveness". Clearly the natural place to start. For lc this seems to be referring to:

(1) exclusively physical looks sans personality, clothing, power, make-up, presentation, social position &c. 

(2) the assessment of said looks by men and women, according to the metric of 'sexiness.' 

Let's shelve for a moment the possibility of women mistaking what men think is sexy (creating what we could call 'pseudo-sexiness'). 

Let's simplify and assume it is the sex appeal of women to heterosexual men. By sex appeal, let's take it as "desire to have sex with said women"  and not merely "assessment of sexiness."

Research direction 1: Wild men

A problem: human sex appeal, as lc recognises, is clearly different from animal sex appeal. It has a layer of symbolic representation. 

Bodies can be sexualised in different ways in different periods depending on media, power, history, art and so on. E.g. eras of skinny women and curvy women in cycles, or, more weakly, the photoshop "make this woman beautiful" test. In the 19th century womanly moustaches were considered attractive:

One solution to this problem is to take a massive sample of female bodies across history and present them (in the form of holograms, sex-bots, or very realistic body pillows) before a large sample of grown-up feral children (henceforth: wild men) from a variety of genetic and climatic backgrounds. Do a large number of pair-wise comparisons; record which of the women the wild men prefer sexually and by how much. Control for wild man horniness and other variables. Possible outcomes could be: 

1. All men, regardless of characteristic (e.g., climatic conditions, being raised by wolves or birds, genetic profile) align on preferences. From this we can produce a distribution curve. Call this the "Wild-man biological default" (WMBD).
2. Men differ in preference depending on characteristic, producing clusters of curves. Call these "cluster defaults."
3. There is little to no inter-coder reliability between wild-men. Sexuality is entirely idiosyncratic. Cf. this discussion of beauty by Gwern citing a study suggesting high inter-coder reliability for beauty. 

This is probably as close as you can get to a pre-cultural selection process. If I had to bet money on 1 and 2, I'd predict an inegalitarian bell-curve distribution on the basis sexiness depends on a number of body parts which can all vary in symmetry, healthiness, shape, colour, &c. That said, there is a possibility they might be so feral as to lack the ability to make finely-graded distinctions. 80-20 for inegalitarian. 

I would be genuinely curious on the ur-sexy woman, although contrary to lc's photo examples I doubt facial structure would be particularly weight-bearing. 

Research Direction 2: messy cultural experiments

Another approach is to survey modern non-feral men. Existing studies typically focus on waist-hip ratio (WHR) and volume-height index (VHI). Some options are:  

1. Men in diverse cultures throughout history, are straightforwardly (if sometimes secretly) attracted to the same features. These features might be the same as the WMBD/cluster default.
2. Men in diverse cultures may initially differ, but will converge on the same features under conditions of unlimited free choice, availability, and low social pressure for conformity. One night with WMBD is enough.
3. Men, through modernity, achieves a higher order of consciousness (e.g. via  Hegelian dialectic) and discovers a superior form of the WMBD, which we will call the Free Man Biological Default.
4. Female sexiness is a social construct all the way down. WMBD does not exist; your assessments of hotness are not 'natural' but learned. Any existing studies are biased by the pervasive conditions of westernisation, capitalism, and modernisation. Expect massive cultural divergence otherwise. 

It depends on the features which are attractive for 1 and 2. Nonetheless, if I had to bet money, I would also predict a bell-curve. I predict the same for 4 given social dynamics, as do cultural theorists who criticise hierarchies of sexual desirability. 

Conclusion

Absent a deeper review of the data, I suspect sexiness is not particularly egalitarian. Women probably exist across a wide range of sexual attractiveness. 

The upper-end might not be super long - Sofia Loren and Bela Lugosi might score close to one another - but I suspect even abstracted from their fame (e.g., presented to a variety of uncontacted, non-homosexual tribesmen) they would be significantly ahead of the modern average woman, who is, in turn, yet more vastly ahead of the pre-modern fish-woman.