Abstract Much of the work on transmission of solar radiation in solar ponds is, like that in the oceans, based on measurements made in so called “pure water”, that is, either distilled water or clean sea water which has been extremely well filtered. The original calculations were published in 1908, based on transmission measurements for distilled water published in 1895 and using the original Langley spectrum dating from 1882/83. Yet a solar pond may contain 20% or more of sodium or magnesium or magnesium chloride, or a mixture. Whilst unfiltered solutions were known to exhibit very large transmission loss, it is now found that extreme filtering again achieves the transmission of “pure” water. Moreover, prolonged settling has that same effect, thus explaining why, yet again, salt gradient solar ponds work so remarkably well.