The per cent of loss in the impact and abrasion test will depend upon the care employed in culling the brick and in select ing the samples, as well as upon the character of the brick. To show the results that may be expected, the following data ob tained by a city in Ohio in the ordinary course of business are given. The samples were selected from material after delivery upon the street, by a representative of the city. The tests were carefully made according to the N. B. M. A. standard as above. The material was in the form of blocks approximately 3" X 4" X 9". The average of the losses of fourteen samples was 18.97 per cent, the range being from 15.60 to 24.35 per cent. The samples repre sent some of the best paving material in Ohio. Omitting the pro duct of one manufacturer, the average of the losses was 18.57 per cent, and the range was from 15.60 to 21.62 per cent, the last being the loss of a well-known standard paving block. Of the product of the thirteen manufacturers, nine had a loss of less than 20 per cent, seven less than 19 per cent, four less than 18 per cent, and two less than 16 per cent. Eleven lots of the best grade of blocks gave an average loss of 15.60 per cent with a range for the average of duplicate tests from 12.25 to 18.70 per cent. Individual tests of these blocks ran as low as 10.3 per cent, and duplicate tests of samples selected by the manufacturer (not included above) gave losses of 11.76 and 11.92 per cent respectively. Ten samples of the blocks having the greatest losses gave an average of 21.62 per cent with a range from 14.90 to 34.77; and samples selected by the manufacturer (not included above) gave losses of 11.14 and 13.73 per cent respectively.
Ten lots of blocks tested in an lllinois city gave an average loss of 18.34 per cent with a range from 15.4 to 24.6 per cent; and, omitting the largest result, the average was 17.64 per cent with a range from 15.4 to 21.2 per cent. Of the ten kinds of blocks, two had losses of less than 16 per cent, four less than 18, six less than 19, and eight less than 20 per cent.
A few other scattering results seem to show that the above are fairly representative of the above localities; but these localities are favored in native material suitable for making paving brick and also in the attention given to that industry, and consequently other localities may not expect as favorable results. The Second Annual Report on the Highways of Maryland contains, on pages 118 to 120, the results of one hundred and twenty N. B. M. A. rattler tests on paving bricks, some of the samples having been selected by the City Engineer of Baltimore and some by the man ufacturers. The results by localities are as follows, in the order of the quality of the bricks: Eleven lots from Ohio gave a mean loss of 18.2 per cent with a range from 16.2 to 20.0; ten lots from
West Virginia had a mean loss of 22.7 per cent and a range from 17.0 to 35; twenty-four lots from Pennsylvania showed a mean loss of 26.8 per cent and a range from 18.6 to 55.8, and omitting two lots the average is 24.8 per cent and the range from 18.6 to 34.2; and nine lots from Maryland gave an average loss of 32.1 per cent with a range from 25.3 to 48.7, and omitting one result the mean is 26.3 per cent and the range from 25.3 to 37.6.
All of the above data are for blocks approximately 3" X 4" X 9". Bricks approximately 2" X 4" X 8" will lose from 2 to 6 per cent more than the above blocks; but not enough data have been accumulated to determine with any accuracy the effect of size upon the loss in the rattler test.* It is difficult to secure the specimens necessary in making the comparison.
A study of the details of the experiments referred to above, indicates that an occasional manufacturer can furnish pav ing blocks giving a loss of 15 per cent or even less; but whether it is wise so to specify will depend upon the service required and upon the cost of different grades of paving blocks. A severe specification will require more careful culling of the product of the kiln and will also limit competition,—both of which demands will increase the cost. The limit to be specified in any particular case will depend upon the special conditions.
The N. B. M. A. standard rattler test is defective in that it determines only the average loss of two charges of brick and gives no information as to the uniformity of the quality of the individual bricks. Uniformity of wear is nearly as important as durability, for a single soft brick soon causes a hole, and the blows of the wheels in dropping into this hole soon destroy the adjacent pavement, however the brick are. The value of the test would be materially increased, if in specifying the limit for the average loss in the rattler a statement were also made of the amount that the loss of one charge may be permitted to go above or below the average of two. A study of the tests described in the preceding section shows that loss of each charge should not vary from the mean of the two by more than 10 or 12 per cent of that result. This modification would add nothing to the cost of the test or to the time required in making it.
The value of the test could be further increased by detcrmin ing the loss of each individual brick, and specifying a limit to the variation of any .brick from the average of the charge. Exper iments show that with the best paving blocks the loss of any indi vidual brick will differ 25 to 30 per cent of the mean of the charge from that result. Part of this variation is due doubtless to acci dental differences in making the tests, but a large part of it is due to lack of uniformity of the bricks themselves.