The results of some tests made under different conditions are here given:

A fire and water test on a concrete expanded metal floor composed of 6 1/2 inches of concrete mixed in the proportion of 1 part Portland cement, 2 parts sand, and 5 parts cinders, showed the following results:

The slab was of a type similar to that shown in Fig. 55, the beams being 20-inch beams and spaced about 12 feet center to center, with the span of the beams about 17 feet 9 inches.

The slab of concrete was loaded with 400 lbs. per square foot, under which it deflected .30 inch. Under exposure to fire the deflection increased to 2 1/8 inches, and when the test was completed remained about 3 5/8 inches.

A portion of the under side of the concrete vas knocked off by the stream of water.

A test under practically the same conditions as above was made of the Columbian system. The type was of the general form shown by Figs. 65 to 67. The spans were the same as above. The slab was 8 1/4 inches in depth, composed of 1 part Portland cement, 2 1/2 parts sand, and 5 parts broken stone. The bars were 5-inch bars, spaced 2 feet center to center, and fastened to the beams by angles.

A portion of the girder covering consisted of mackite blocks plastered, another portion consisting of 2-inch cinder concrete, the latter being the regular construction. There were also two 8-inch I-beams set up, one covered with cinder concrete to 9 inches X 13 inches; the other covered with hollow bricks to 12 inches X 16 inches, giving 4 inches covering.

The floor was first loaded with 1,000 lbs. per square foot, under which it deflected 3/8 inch. The load was then reduced to 400 lbs. per square foot, and the fire test commenced. This lasted for two and one-quarter hours at a maximum temperature of 1,700°. A stream of water was then applied for 4 1/2 minutes, and afterwards another fire test given of 38 minutes and a second stream of water applied. The floor, at the end of the test, showed a deflection of 1 1/8 inches. The cinder concrete beam and column coverings were -not materially damaged. The mackite covering was entirely stripped off, and the hollow brick column covering badly damaged. No apparent injury occurred in the floor slab. After this test the floor was loaded up to 1,650 lbs. per square foot, at which point the walls of the test house made it necessary to stop. The net deflection under this load was 1| inches. A few cracks appeared in the ceiling under this load, most of them being parallel to the bars.*

Numerous other tests of expanded metal floors on shorter spans have shown satisfactory results. For spans up to 8 feet and loads under 200 lbs. per square foot, which are the ordinary conditions, the cinder concrete shows safe results. Beyond these limits special tests should be made in each case.

*Note. A detailed report of these tests is given in Engineering News, June 27, and November 21, 1901.

A valuable review of the effects of a practical fire test on terra cotta and concrete floor construction, is given in the discussion bearing on the fire that occurred in the Home Building, Pittsburg, Penn., May 3, 1897, which was published, in Engineering News, May 20 and 27, and July 1 and 15, in that year. An account of a second fire which occurred on April 7, 1900, is published in the same periodical under dates of April 12 and April 26, 1900.

The New York Building Department conducted a test on three arches of the Gruastavino type, each 3 feet in length. The spans were 6 feet, 10 feet, and 12 feet. The 6-foot span was composed of 2 courses of tile, making a thickness of 2 1/2 inches; the 10-foot span, of four courses, giving 5 inches thickness; and the 12-foot span, of three courses, with a total thickness of 3 3/4 inches. All were leveled up with concrete. The 6-foot span carried 2,500 lbs. per square foot, and showed a maximum deflection of .13 inch. The 10-foot span carried 3,600 lbs. per square foot, with a deflection of .19 inch. The 12-foot span carried 3,125 lbs.per square foot, with a maximum deflection of .32 inch.

This was a simple loading test with no application of fire and water.

Tests of porous terra cotta hollow tile arches have not been so numerous, especially under fire exposure. Table XIII gives the results of a series of tests to determine breaking loads of different arches, and is taken from the " Transactions " of the American Society of Civil Engineers, Nos. XXXIV and XXXV, of 1895 and 1896.

In terra cotta arches as in concrete arches, great variations in strength will result from varying degrees of thoroughness in construction. These arches should always be set in cement and carefully keyed, and the use of broken blocks should be avoided-Settlement in arches of this type often results in cracks in tile or mosaic floors.

Table XIII. Breaking Loads Of Hollow Tile Arches

In the above table the following abbreviations are used: "E" - end construction; "S" - side construction; "H" - bard clay: "Porous" - porous terra cotta; "Dis." - distributed load; "Cen." - concentrated load at center; "Port." - Portland cement; "N. M." - no mortar.

The loads per square foot in the above table were obtained by dividin. the total load by the superficial area of the arch in square feet. The hori zontal thrusts were obtained by the regular formulae; for central loads these are double the thrusts for distributed loads of the same weight.