28 - of the cavities conform to the arrangement of the circular magnesite patches amongst which they lie. The impression gained is that the magnesite patches have grown upwards and sideways forcing the boulders to one side and finally rising above them. That this is the true explanation is further indicated by the presence of slight overlaps of the magnesite on individual boulders. On the borders of large deposits outcropping at the surface, there seems in places to be an overlap of soil resting on the magnesite; that is, in places, excavations made just outside of the main deposit reveal a foot or so of soil resting on a thin bed of hydromagnesite evidently a project- ing part of the main mass. _ : ¥, At Kelly lake and at Clinton near the main masses of hydromagnesite are occurrences of impure earths, carbonates and sulphates of magnesium and caleium, which form nodules lying at or just under the surface or else occur as beds interstratified with sand or clay. One such nodule has been exposed in pit No. 7 (Figure 3), at Clinton (Plate V, and analyses 7 and 8, Table IV). It is overlain by 6 inches of black soil and consists of nearly pure gypsum and calcite along its upper border, but in depth these minerals are mixed with gradually increasing amounts of the boulder clay ~ in which the nodule lies. Under the nodule, traversing the boulder clay, are fine, thread-like streaks of cream-coloured earth identical in composition - with the materials of the nodule (Plate VI). Nodular deposits of gypsum also occur on the steep hillsides at Kelly lake (Table IV, analysis 1). 2 Bedded deposits of calcite admixed with some gypsum and hydrous mag- nesium carbonate (Table IV, analysis 2), are present in the shallow parts of Kelly lake where they lie within a few inches of the surface of the water (Plate VII) and deposits of the same character (Table IV, analysis 3) occur interstratified with sand and clay in the flat west of Kelly lake. COMFOSITION. In the following table, analyses are given of the better grades of hydromagnesite found at Clinton, Meadow lake, Watson lake, and Riske creek. This table represents the composition of the material -in these deposits that may reasonably be expected to yield a commercial product. The yellow, granulated material under it is too high in lime as a rule to be of value. Analyses 1, 2, 4, 6, and 7 are of samples taken in such a way that they represent an average of the full thickness of the layer sampled and as stated in the table, in most instances the sampled layer represents the whole of the upper white bed. With these analyses is presented the average value of eight analyses of Atlin hydromagnesites taken from the report by G. A. Young in the Summary Report of the Geological Survey, Canada, for 1915 and also the average of sixteen analyses of California magnesites. Analysis 5 was made in the Geological Survey laboratory more than twenty years ago, on a sample which it is presumed was taken from the cleanest material found near the surface of the deposit, but partic- ulars as to the method of taking this sample are lacking. These results show the materials to be low in lime and other impurities. Analyses 2 and 4 reveal the presence of 4 and 4-6 per cent of silica respectively, which in all probability is derived from the fine, black specks scattered through the upper white hydromagnesite in many places. The composition of the spring waters (page 40) indicates that hydrated amorphous silica also may be present. The magnesia content runs from 41 to 43 per cent.