Out of interest, are the v-shaped all-the-way-to-zero drops at 393½ and 397 nm real or somehow an artifact of the processing? The whole thing looks intuitively like the absolute value of an amplitude that drops below the axis and climbs back up again, but I have trouble imagining a physical effect that would cause it to look like that.

@HenningMakholm The shape of the atmospheric O2 (Fraunhofer) lines looks "distorted" because there is a pressure gradient in earths atmosphere. So one observes a superposition of many Lorentzian/Voigt lines of different width. Because line width is approximately proportional to pressure it creates a more triangular shaped absorption profile for the atmospheric lines.

I apologise if this is too divergent and perhaps should be a separate question, but I saw the close-ups and thought "there are so many lines/dips!". Presumably every single one corresponds to elemental absorption..? Then I thought: in school we're taught that the sun's emission spectrum is continuous (ignoring the fraunhofer lines, et al). But is it really? Or is it just the sum of very VERY many emission lines of various elements in the sun. Emissions should be quantised to electron shell transitions, and now I can't see why anything would be truly continuous.

@Barney It is a different question. A very interesting one, but you really should ask it separately. But in short: (1) lines are never infinitely thin because the emitted wave packets have a finite length (“lifetime=uncertainty broadening”): that’s why particle physicists use “decay width” more or less interchangeably with “state lifetime”; (2) at finite temperature and density, this ludicrously small width is increased by several orders of magnitude, mostly by “thermal=Doppler broadening” and “pressure=collisional broadening” in a gas; (3) most importantly, in an equilibrium setting (such as that of blackbody radiation), it doesn’t matter whether the absorption at a particular energy is strong or weak as long as absorbing layer is thick enough: it doesn’t matter whether the photon got absorbed on the first or on the billionth encounter as long as it didn’t come out; thus those apparently insignificant tails of the spectral lines end up being as important as the peaks.