When these experiments were first commenced we were still under the impression, which several months' experience of working with gums had produced, namely, that the Ghattis were quite distinct in their properties to ordinary gum arabics. But the new hypothesis, and the experiments undertaken to confirm it, showed clearly that if the viscosity of a gum solution depends on the ratio of metarabin to arabin, then there is no absolute line of demarkation between a Ghatti and a gum arabic. In other words, there is a constant gradation between gum arabic and Ghattis, down to such gums as cherry gum, consisting wholly of metarabin and quite insoluble in water. Therefore those gum arabics which are low in viscosity consist of nearly pure arabin, while as the viscosity increases so does the amount of metarabin, until we come to Ghattis which contain more metarabin than their arabin can hold in solution, when their viscosity goes down again. From these observations it would follow, that by taking a gum of less viscosity than the gum arabic previously used to dissolve the Ghatti, less of it would be required to do the same work. We confirmed this suggestion experimentally by taking another gum arabic of viscosity 0.0557 at 15° C. A mixture containing 93.3 per cent. of this Ghatti and 6.7 per cent. of our thinnest gum arabic gave a clear solution which had the highest viscocity we have yet obtained for a 10 per cent. solution. ----------------------+--------+-------+ H. Pressure 200 mm. | [eta] | Z. | Temperature 15° C. | 0.5525 | 4,456 | ----------------------+--------+-------+ This gum arabic may be regarded as nearly pure arabin (as calcium and potassium, etc., salt). By diluting the new "maximum viscosity" solution, therefore, with the 10 per cent. solution of the gum arabic in fixed proportions we obtain a series of viscosities which are shown in the following curve. [Illustration: Curve Showing Influence of Ghatti upon Viscosity.] Besides obtaining this curve for change in viscosity from maximum amount of metarabin to no metarabin at all, we also traced the decrease in viscosity of the "maximum" solution by dilution with water. The following numbers were thus obtained, and plotted out into a curve. Having obtained this curve, we are now in a position to follow up the hypothesis by calculating the surplus amount of insoluble matter in a Ghatti. For, let it be conceded that the solution of any Ghatti leaving an insoluble residue is a mixture of arabin and metarabin in the same ratio as our "maximum" solution, only more diluted with water, then from the found viscosity we obtain a point on the curve for dilution, which gives the percentage of dissolved matter. Now to show the use of this: The Z value for a 10 per cent. solution of the second Ghatti at 15° C. is 2,940. This corresponds on the curve to 8.4 dissolved matter. 10 - 8.4 = 1.6 grammes in 10 grammes, which is insoluble.