DESIGN OF TEST BAR APPLIANCES, ETC. 517 poured. Unless great care is used, one may be able to make them "run'' fully four inches farther than if they were poured steadily, whereas, when poured vertically, as in the author's system, if there is a quick dash at any time it cannot raise the metal in the fluidity strips any faster than in the test bar moulds, thereby causing a natural and equal rise to truly denote the metal's fluidity or life at the moment the bars are poured. To obtain the contraction of a bar, the distance between the points or tips V V, Fig. 122, page • 514, is measured. These contraction tips are accurately cast in the mould by means of four projections forming part of the flask, two of which are seen at B B, Fig. 123, These projections " chill" one face of the contraction tips V V, thereby giving a clean face to measure from. The lower tips are given form by reason of a swell being made at the base of the fluidity strips, as will be seen at the lower V in Fig. 122, The upper tips are formed by having loose tip patterns placed in the recesses of the mould board as seen, in such, a manner that the uppermost projection B of the flask is on the top side of the tip V. By this arrangement full freedom for expansion at the moment of solidification is permitted, as when this takes place it can extend its length downward in the sand forming the bottom of the mould. These contraction tips are cast twelve inches apart and will be found as arranged to provide positive points for obtaining the contraction of any 'c grade '' of iron. At A, Fig. 122, is seen the pattern used for forming the pouring basin and runner which leads to the 11 whirl-gate." At N is shown how the pouring basin and runner look before being broken from the testsld ask such, after having I''