The abrasion of pebbles rolled in a large concrete basin by a revolving current,
both on a sandy and on a pebbly floor, was studied – a setup believed to be a substantial improvement on the customary tumbling- mill experiments.
One of the most significant results is that, on a sandy floor, abrasion is less than on a pebbly bottom under similar conditions.
The difference increases with size of the rolling pebble and can be four or five times less for medium pebble sizes. Hence reports on measurements of roundness in nature should always be accompanied by statements not only of the size of the pebbles, of current velocities, and which rock type investigated but especially as to the nature of the bottom. On a sandy bottom, weight and velocity have only slight influence on the percentage of abrasion per kilometre.
With increasing roundness, there is a small reduction in the rate of abrasion.
It was found that abrasion on a pebbly floor is reduced by 10-15 per cent by the introduction of sand. The abrasion on a pebbly floor increases in proportion to the square of the velocity. Over the natural range of stream velocities it is doubled for fine gravel. Pebble weight is of even greater importance,
the percentage of loss at low velocities being proportional to the diameter.
It increases three to four times between fine and coarse gravel. Increasing roundness,
from sharp-cornered to sub-rounded, reduces abrasion significantly; beyond sub-roundness, abrasion is reduced much less. As a result of these relations, coarse gravel can loose four to five times as much in percentages as can fine gravel at equal velocities. Twenty-five times as much loss per kilometre is possible under extreme conditions of velocity, size, and bottom cover.
The abrasion processes are shown to be as follows: splitting (= breaking), crushing, chipping, cracking (superficially), and grinding.
Sharp-edged material rolled on a pebbly floor chips during the first 2-10 km. of transport, loosing up to five times as much as by cracking, which is the main process later on.
On a sandy bed, only grinding takes place.
Roundness measurements by Cailleux's method tend to give values that are too high because of parallax, different investigators obtaining different results.
Cailleux's method could be somewhat improved by using the middle instead of the longest axis.
It should be realized that a pebble will pass relatively very swiftly through the first roundness class and still quite fast through the second and third, then more slowly,
until, beyond about the fourth class, the passage from class to class, on an unchanging floor, will require greater and greater distances of rolling until the ultimate shape is attained.

The conclusion is reached that splitting is rarely the result of impact, except in poorly consolidated or fissile rock, but breaking does occur and must be attributed to weathering, especially frost action, on pebbles while they form part of alluvial deposits alongside the stream course.
Poser and Hövermann studied rounding in rivers of the Harz. The present investigation shows that abrasion of the Graywackes they studied is hastened at the start because the solifluction material is superficially weathered.
In the experiments on a pebbly floor the abrasion was about four times as much per kilometer as in nature.
This indicates that in these natural streams much of the bed may be covered with sand.



The rounding, shaping, and loss in weight of pebbles by wear on beaches and in rivers have been studied by a large number of authors. Their particular interests have been varied, as well as their methods of approach. Laboratory studies on the forms and field work on shapes and sizes have been made, mainly by statistical methods. Various systems have been developed for expressing shape or roundness.

To few problems in geology have experimental methods been applied so widely as to the rounding of pebbles. Experimenters invariably used tumbling mills, a rather unsatisfactory procedure. All these studies have greatly increased our insight and have shown this apparently simple problem to be full of complications, most of which are still incompletely understood.

It has also been claimed that studies of roundness should teach us the history of the deposits examined, distance and medium of transport, physiography and climatic conditions, etc.
There are, however, so many variables involved in the rounding of pebbles, including the nature of the bottom, that it appears doubtful whether quite so much of the history of a deposit can be read from its roundness histogram, even if combined with other shape measurements.

In a former paper the writer (Kuenen, 1956) treated the problem of wet sandblast- ing, a process held by some to be highly important for abrasion of pebbles. The conclusion reached by experiment was that sand- blasting is not significant for pebble sizes; for, at the velocity where abrasion begins, any pebble will be rolled along and soon pass beyond the reach of the sandblast. Only boulders can remain stationary in a current of sufficient velocity to transform them into aqua-facts (Kuenen, 1947).

The present article describes an experimental study of pebble rounding during transport by a current of water, a method hitherto neglected. The results are discussed partly to find the nature of the abrading processes, partly in connection with the field work of the French school.

The first to carry out experiments on pebble abrasion was Daubree (1879). He rolled fragments of feldspar and granite in a revolving cylinder under water and showed that they became rounded and that a fine colloidal mud was produced.

Rock fragments were used as abrasives by Wentworth and Krumbein, but Lord Rayleigh had metal fragments as grinding material to abrade his rock pebbles, and Marshall added sand to his pebbles. Schoklitsch allowed single pebbles to roll with sand.

What takes place in a tumbling mill, however, is a very poor imitation of the natural action in rivers, as Bonney (1888) realised long ago. The present author's objections to tumbling mills are the following. The water, instead of impelling the fragments, acts as a brake.
The movement of the fragments in the mill may be either more or less rotatory than in a river. The fragments alternately drop down a steep slope and then lie still until dropping again, instead of rolling continuously over a horizontal surface
The rolling velocity is therefore not distance divided by time. It stands in a complicated relation to the rate of revolution and the amount of sediment involved.
Moreover, the latter has not been determined by the experimenters. The influence of various rolling velocities, that is, the factor controlling the rigour of the process, is therefore known only qualitatively. In experiments with many pebbles these interfere with one another's movements, the front of one touching the rear of its leader with an opposite direction of movement, whereas in nature pebbles nearly always roll separately.
In some ways the experiments in mills are closer replicas of surf action on a pebbly beach than of stream action. Rolling on a sandy bottom has not been imitated successfully.


Because of the drawbacks inherent in the use of tumbling mills, the writer decided to experiment with a revolving current.
The rolling velocity was first measured by using pebbles painted white; counting the number of revolutions and measuring the aver- age diameter of the circle followed. This method was not very satisfactory because the pebbles showed up very poorly, especially at high velocities. Most of the experimental results therefore show an uncertainty of about 10 per cent as to velocity and distance. Luckily, this amount is small compared to variations due to other
Only toward the end of the investigation was a much better technique evolved. This consisted in painting a pebble with fluorescent paint and watching it in ultraviolet light. The water is first treated with KMnO4 to eliminate fluorescing organic matter. With this method, the size of the circle followed could be accurately measured, even for small pebbles at high speeds, and the velocity or travel distance determined to within a few per cent.
Whether on a pebbly floor the addition of sand has any influence was not investigated, because the experiments using these accurate velocities were made on a clean floor. But when sand was added on the cement floor, the circles described were altered. A rough correction was made by feeling under water with a piece of stout wire to determine the size of the circle described by the rolling pebbles. Obviously, this is not a very satisfactory velocity measurement, although the results were adequate for first approximations.


Lord Rayleigh
W. C. Krumbein