Abstract

The two types of abrasion testing machines which gave the best correlation with wear of shoe soles in previous wear trials1were used in further experiments with other and more numerous leathers. On thirteen leathers three independent wear trials were made in different countries, with precautions taken to make these trials as reliable as possible. Two pairs of shoes were worn by each wearer, with reverse soling (leathers A and B, say, as left and right soles of one pair, and B and A as left and right of the other). The soles compared were taken front the same bend location, were attached to shoes which were all of the same construction, and the two pairs were worn on alternate days. The relative durabilities of the leathers were measured in various ways, and attempts were aede to apply corrections of various kinds: for compression of the soles in wear, for baching of water soluble material, and for the different durabilities in each layer of leather from the grain to the u1sli. The results differ slightly according to the method of measuring durability, and according to the corrections applied. (The corrections are plausible, but their accuracy is doubtful.) Whatever the method and the corrections, however, the agree ment between the three wear trials was extremely good, and it is estimated that if further sets of trials like the present ones were made, the correlation coefficients between sets of trials would lie between +096 and – 098 for town wear in temperate climates, despite the usual differences of weather and pavements. l-Ience, if an abrasion test were perfect, it might be expected to lead to a similar correlation coefficient when compared with the present series of wear trials. None of the abrasion testing machines gave as good correlation with wear as this, but two out of three gave correlation coefficients of about +090 (or somewhat higher when used in an appropriate manner with an appropriate pressure between the specimens and abrasive. The optimum pressure is higher for the Dutch abrasion machine2which uses sand as abrasive, and is much less critical than for machines which use an abrasive cloth. The Dutch machine gives good correlation with wear, whether the abrasion tests are made on dry or wet specimens, whereas the other type gives high correlation coefficients with dry, but not wet specimens. There is not much evidence with either machine that it is worth making wet as well as dry tests for prediction purposes. Both types of machine are capable of giving fairly accurate prediction of wear from abrasion tests on a few specimens, hut the Dutch machine is superior in this respect.