discussion of eclipse observations were true, Dr. Downing, the Superin- tendent of the ' Nautical Almanac,' having been good enough to write out to us in Sumatra about this result of the discussion. It was arranged that Hobart should continue the counting of seconds after totality was finished, so that I might note on the chronometer the instant at which the final " sixty " was called.
The operations were rehearsed several times on the two days before the eclipse, and everything went off as perfectly as rehearsals could make it. In a later paragraph reference will be made to one unforeseen mishap which arose from the interference of the clouds.
7. The Day of the Eclipse.
When the sun rose and the morning mists were dispelled on the 18th the day of the eclipse the conditions seemed vastly more favourable than on any day in the previous week, during which it must be confessed we had rather a trying time, inasmuch as sunshine was needed for the final adjustment of the two spectroscopes, and it was very difficult to find moments between clouds when enough could be got.
The early hours of the morning of the 18th were occupied in com- pleting the adjustments, which, as it turned out, were already nearly perfect, except in the case of the quartz spectrograph, as will appear in a later paragraph. The clocks for driving the ccelostat and the large spectrograph were finally rated, and we had gathered for a last rehearsal, more for the sake of establishing confidence than of adding to the instruction and drill, which had been gone through many times on the two previous days. But about 1 1 o'clock, very soon after I had observed the first contact, clouds began to move up from the N.E., and to form round the sun, and it began to be an anxious time. I was con- tent to dispense with the rehearsal in dumb-show, and thus avoided the risk of inadvertent exposure of the photographic plates in the dark slides, preferring to give some finishing touches to the arrangements for signals, exposures, and observations. Nothing was lost by this procedure, and in the actual event the programme was perfectly carried out by everyone concerned, except that cloud interfered in an unfore- seen way (see 11, p 221) with my operations with the 4-prism spectro- graph, thus curtailing my most important exposure by 2 minutes, and also involving the loss of one of the exposures with the double tube.
Fifteen minutes before totality the cloud was still hanging about, and obviously forming in some places and dispersing in others. Iris- coloured rings were seen round the diminishing crescent, showing that the clouds were water-clouds, not ice-cirrus. Ten minutes before totality I looked into one of the spectrographic instruments (see
