water-colour work, for when a wash of pigment is passed over it, the colouring matter and the water partially separate, while the outline of the brush-stroke is not preserved. Before entering further into the question of what are the essential and what the accidental and unnecessary constituents of paper, I give the summarized results of six analyses, which show the percentage proportions found in good samples :
| Water | Size | Ash | Fibre | |
| Hodgkinson, 1869 | 6.8 | 4 | 1.1 | 87.5 |
| English, 1876 | 10.9 | 6.1 | 1.1 | 81.9 |
| Dutch, 1876 | 11.0 | 4.8 | 0.9 | 83.3 |
| Whatman. 1885 | 7.4 | 6.3 | 1.1 | 85.2 |
| Arnold, 1894 | 7.4 | 7.6 | 1.5 | 83.5 |
| 'O. W.,'1897 | 8.7 | 5.5 | 1.7 | 84.1 |
Water.— It should be noted that the percentages of water shown in these analyses vary considerably by reason of variations in the humidity, temperature, and pressure of the atmosphere to which the different papers had been exposed just before the analyses were made. There are, however, slight permanent peculiarities in samples made from different fibres or sized in different ways; in consequence the moisture-absorbing and moisture-retaining properties of different papers are not precisely identical under identical atmospheric conditions. This hygroscopic moisture does, indeed, vary inversely with the temperature, and directly with the amount of water-vapour in the air; it is increased also by an increased barometric pressure. There is no doubt that if it could be wholly excluded, the larger number of changes which occur in the pigments of a water-colour drawing would be prevented. It is most injuriously active when a framed drawing is exposed to considerable
