His experiments were begun towards the end of May, 1882, and by September 28 he had obtained a fair earnest of success. The exclusion of all other qualities of light save that with which he desired to operate, was accomplished by using chloride of silver as his sensitive material, that substance being chemically inert to all other but those precise rays in which the corona has the advantage.[544] Plates thus sensitised received impressions which it was hardly possible to regard as spurious. "Not only the general features," Captain Abney affirmed,[545] "are the same, but details, such as rifts and streamers, have the same position and form." It was found, moreover, that the corona photographed during the total eclipse of May 6, 1883, was intermediate in shape between the coronas photographed by Sir William Huggins before and after that event, each picture taking its proper place in a series of progressive modifications highly interesting in themselves, and full of promise for the value of the method employed to record them.[546] But experiments on the subject were singularly interrupted. The volcanic explosion in the Straits of Sunda in August, 1883, brought to astronomers a peculiarly unwelcome addition to their difficulties. The magnificent sunglows due to the diffractive effects on light of the vapours and fine dust flung in vast volumes into the air, and rapidly diffused all round the globe, betokened an atmospheric condition of all others the most prejudicial to delicate researches in the solar vicinity. The filmy coronal forms, accordingly, which had been hopefully traced on the Tulse Hill plates ceased to appear there; nor were any substantially better results obtained by Mr. C. Ray Woods, in the purer air either of the Riffel or the Cape of Good Hope, during the three ensuing years. Moreover, attempts to obtain coronal photographs during the partial phases of the eclipse of August 29, 1886, completely failed. No part of the lunar globe became visible in relief against circumfluous solar radiance on any of the plates exposed at Grenada; and what vestiges of "structure" there were, came out almost better _upon_ the moon than _beside_ her, thus stamping themselves at once as of atmospheric origin. That the effect sought is a perfectly possible one is proved by the distinct appearance of the moon projected on the corona, in photographs of the partially eclipsed sun in 1858, 1889, and 1890, and very notably in 1898 and 1900.[547]