In fact, in the article “Eye protective techniques for bright light,” published in Ophthalmology 90, 937-944 (1983), David H. Sliney wrote: When the sun is low in the sky it is yellow or orange indicating that the hazardous blue light has been scattered out of the direct path of sunlight, and the sun may be fixated for many minutes without risk. It's worth going through the numbers for this situation, because there is a very large and rapid change in the brightness of the Sun near sunset. For example, the smallest possible atmospheric extinction coefficient at sea level in blue light is about 1/4 stellar magnitude per airmass (the airmass at the zenith is taken to be unity.) When the Sun is 5° above the astronomical horizon, the airmass is about 10, so the blue light is reduced by at least 2.5 magnitudes, or a factor of 10. This would ordinarily not permit a threshold lesion to develop in 100 seconds; if we suppose the damage depends only on the total exposure, then 1000 seconds would be required, assuming the brightness remained constant. But, at low latitudes, the Sun sets 20 minutes or only 1200 seconds after reaching an altitude of 5° — and during this time, its brightness is rapidly decreasing. This suggests that, at low latitudes, staring at the Sun for the full 20 minutes before sunset might be marginally enough to produce a threshold photochemical retinal lesion in an average eye. As there is evidently some variation in sensitivity, not all eyes would necessarily be safe at this point. A prudent observer might ask for an additional factor of 10 to be safe. This requires waiting until the Sun reaches 20 airmasses, or about 2° altitude, 8 minutes before sunset at the Equator, or 12 minutes before sunset in places like Montreal, Paris, or Rome. At higher latitudes, the Sun is lower and even safer to look at 10 minutes before sunset; so “10 minutes before sunset” seems a safe rule to employ. As the width of the thumb at arm's length is just about 2°, it is a good “rule of thumb” that if you can cover up the image of the Sun with your thumb, extended at arm's length, and still have the lower edge of the thumb touching the sea horizon, you can look at the Sun safely. A very conservative observer who wanted the full factor of 1000 attenuation of blue sunlight recommended by Ham et al. would wait until the Sun reached 30 airmasses, at an altitude of a little less than a degree (i.e., 2 solar diameters). At this point, “continuous” viewing is safe; but the Sun remains in sight for only 4 more minutes at low latitudes. A more realistic calculation would allow for the additional attenuation by aerosols, which can be quite strong at the low altitudes mentioned here. In fact, the Sun is so attenuated at short wavelengths that the first people who tried to photograph sunset phenomena were continually frustrated by their inability to record an image of the Sun at the horizon on unsensitized photographic plates: see, for example, the paper by Riccò in the bibliography. The short, photochemically active wavelengths required for photography on unsensitized plates are the same ones responsible for photochemical retinal injury; if the setting Sun cannot be photographed at these wavelengths, it cannot possibly cause retinal injury. After I did the calculations described above, I found that similar calculations had been made by D. Sliney and H. Wolbarsht Safety with Lasers and Other Optical Sources (Plenum, New York, 1980) On pp. 205-206, they say: As sunset approaches, the relative fraction of blue light in this direct solar spectrum dramatically decreases as the sun nears the horizon. … [O]nce the total irradiance falls below 3 mW/cm2 (corresponding to an elevation angle of less than 5° at sunset in relatively clear weather), most people find it reasonably comfortable to look at a sunset which lasts for less than 10 minutes. … [They then go through a detailed calculation that need not be repeated here.] This would also explain why an individual who drives toward the sun at low elevation angles as he goes to and from work does not receive a retinal injury. So, when the Sun is touching the sea horizon, it is certainly completely safe to look at. This is in accord with the experience of millions of people who have watched many seaside sunsets without harm.