Some washes of water-colour, of thirty-one different kinds, were made upon cards by the late Mr. W. Simpson. He so cut the cards as to divide each coloured strip in half; one section was preserved in darkness, the other was exposed in an eastern aspect on the shutter of a house in London for fifteen years, but the sun did not shine upon the specimens after ten o'clock in the morning. As they were not tightly framed, the cards became a good deal discoloured by the absorption of noxious vapours and dirt. The results were:

It will be noted that the above results are for the most part in agreement with those recorded by other experimenters; the chief exceptions are marked with a star. Vermilion is usually blackened, but it is possible that the sample employed in these experiments was the less changeable native form or cinnabar. The Vandyke brown, too, was probably the earthy rather than the bituminous variety; the slightness of the change recorded for madder brown and sepia, and the absence of any alteration on the part of bistre, are less easy of explanation. The madder pigments seem to have stood more than usually well, but they often exhibit large differences of stability. Nor must it be forgotten, in assigning values to the above results, that this trial of fifteen years' exposure was not of the severest kind. Although, on the one hand, there was the imperfect exclusion of an injurious London atmosphere, on the other hand, the energy of the solar radiation was much reduced by the prevalent condition of the smoky air, while the intermittent and capricious sunshine of the Metropolis never fell on the trial cards after ten a.m.

The late Mr. R. H. Soden-Smith kindly placed at my disposal a large number of specimens of old water-colour cakes and of powder colours intended for oil-painting. One set consists now of eleven cakes or fragments of cakes (in their original box) bought about the year 1815 of Newman, in Soho Square. This set is peculiarly interesting as the colours, which all bear the name of the maker and his device, represent those used by many of the best English water-colour painters during the first quarter of the nineteenth century. The cakes are: Indian yellow, raw sienna, raw umber, burnt sienna, burnt umber, vermilion, carmine, burnt carmine, pink madder, ultramarine, indigo; neutral tint and sepia are missing. On comparing the hues of the first nine of these paints and of the indigo with the hues of the corresponding cake-colours as sold by the same house in 1886, no appreciable differences were detected save in the case of the raw umber. Here the pigment of 1815 showed a more beautiful nuance than that of 1886. On making comparative tests of the stability, under exposure to sunshine, of the two sets of pigments, the results were found to be practically identical.

One cannot, therefore, claim for the water-colour paints in use one hundred years ago a degree of permanence greater than that possessed by their representatives of to-day.

By far the most important series of trials of water-colour paints yet published is that to be found in the report by the late Dr. Russell and Captain (now Sir W.) Abney to the Science and Art Department (1888). The reporters endeavoured to give precision to their experiments and their conclusions by a careful comparison of the effective radiation from different sources of light. The first part of their report contains a very useful discussion of the relative values of direct sunlight, light from clouds, and from an overcast or clear sky, and light from artificial sources. Several cognate subjects are also discussed therein, such as the number of years of exposure which pigments would require, if in the picture galleries of South Kensington, in order that they might suffer the same changes as those caused by three or twenty-two months' exposure in a southern aspect outside the Museum. Part II. of the report contains the results of twelve sets of experiments with various pigments. In all the series the same paper (Whatman's) was used.

In the paragraph relating to this subject there is, however, one curious error, and one obscure statement (p. 27). It is quite impossible that the paper used - its weight per ream is not given - could have contained so little as 'nearly 1 grain' of sizing matter per square foot; 10 grains is a more probable quantity. The sentences next following do not state the condition of the papers which absorbed from a moist atmosphere from 12.07 to 12.46 per cent. 'of their weight of water.' Were they dried previously, and, if so, at what temperature? We ought to have been told within what limits the percentage of water in these papers varied during the course of the trials: I have pointed out for many years past the importance of this hygroscopic moisture in paper in reference to the fading of pigments-Eight tints of each pigment were applied to strips of paper 8 inches long by 2 inches wide; they were exposed in tubes open at both ends, but having the upper extremity curved downwards so as to exclude wet and dirt.