VAU'K <)K Mli'KOMK.TKK M KASl < l< KM MN TS. 479 It is impossible to obtain roui^h bars of the same area. There is sure to be some difference in their sizes. It is not unusual to find one-inch area, etc., bars to be from one-sixteenth to one-eighth larger in diameter or the square than the pattern used and to find that testers make no note of such difference, but are wholly guided by the weight at which the bur broke. If one was one hundred or two hundred more than others, the highest was accepted as the strongest and best test, regardless of the bar's exact area. To illustrate how a small bur breaking with a heavier load than the lur^e bur (each differing* but a few thousandths of an inch in their area), may often, if not reduced to relative strengths, etc., deceive a tester 200 to 400 pounds in acceptiny; common rule measurement and tlie actual load in thinking he has a true record of the iron's strength, the reader is referred to Table 89, tests Xos. 6 and <S, on pu^'c 460, showing transverse tests of ^un metal. There we find two bars which, if the actual breaking loads were accepted, would deceive the tester .\(n) pounds, or in other words, instead of his believing lie had one bar only 72 pounds stronger than the other, he actually had ;i difference of .j6() pounds, us stated above. Tin's should aid to clearly illustrate the importance of micrometer measurements, wherever the tester desires to truly ascertain whether any difference actually exists in the* strength of his mixtu/es or the character of the iron produced. Another feature well to be noticed is I hut. of the impructibility of obtaining burs exactly round or square, or exact duplicul.es of their pattern. Man)1 testers take but one measurement of a bar, while others take no measurement at. all. Anv following either in a letter discussing the testing of cast iron and attacked the usual formulae for loaded beams asound burs..........................*fi$7 "