Detailed lithologic and grain-size analyses of 174 samples collected along 12 rivers in Alberta were carried out by L. B. Halferdahl. This set of data is used to illustrate downstream changes in the bed material of these rivers. Each of the major rivers can be sub-divided into three reaches on the basis of downstream change in grain-size. The Mountain reaches show highly variable grain-size characteristics with a tendency to increasing size downstream. Through the Foothills and Western Plains plots of grain-size, D, against distance, x, follow the relationship D = D₀e ^ax Where D₀ is the grain-size at x = 0 and a is a diminution coefficient. D and D₀ are given as D₅₀ or D₉₀ obtained from cumulative grain-size curves. Bed material is predominantly gravel in the above two reaches. At the beginning of the third reach the rivers become sand bedded and there is no significant decrease in grain-size with distance downstream.

The reach characteristics can be explained qualitatively in terms of their geomorphic history. The Mountain reaches include numerous lakes and infilled lakes of glacial origin. These act as sediment traps and cause the burial of coarse material. In addition much of the present supply is from periglacial environments in which the bedrock is heavily fractured. The Foothills and Western Plains reaches have been degrading since deglaciation and are expected to show grain-size changes according to Sternberg’s relationship. The Eastern Plains reaches were affected by isostatic adjustment to the Wisconsin, Laurentide Ice Sheet. Simplifying assumptions of the isostatic effect on stream slopes leads to the conclusion that, following initial degradation, these streams should now be aggrading. Aggradation accounts for the abrupt change from gravel-bedded to sand-bedded streams.

Abrasion coefficients from laboratory experiments and diminution coefficients from rivers are compared. Laboratory experiments greatly underestimate diminution coefficients for rivers, and aggrading rivers show higher diminution coefficients than degrading rivers. The diminution coefficient, a, can be divided into three components: a_t, the component of abrasion during transport; a_v, the component of in situ abrasion: and B, the component of differential transport. The value of the diminution coefficient for limestone is obtained using the bed material samples, and also using the diminution coefficient for quartzite and an experimentally derived ratio of the abrasion coefficients for quartzite and limestone. The second derivation of the diminution coefficient for limestone includes an assumption that B = 0. As the two coefficients are similar this assumption is accepted, and the ratio of abrasion in transport to abrasion in situ is calculated for quartzite as

This result explains why laboratory experiments, which do not include a_v, greatly underestimate abrasion coefficients in rivers.