against the faces of the pier by clamping-bolts passed through both of them, two at either end. The steel H-girders, 6 inches by

5 inches by

inch in cross-section, were then bolted underneath and across the ends of the two channels; and screw-jacks were set

on the floor under their centres, as shown in

Figs. 4. The jacks were screwed up until it was judged they were sufficiently tight to prevent sagging. That part of the pier below the level of the channels was then cut away, and a piece of bitumen-sheeting was laid on a bed made even with mortar. The pier-foot was then. rebuilt with bricks and 3-to-1 cement-mortar, underpinning the supported column. The bitumen sheet was of such a size as to project 12 inches all round the finished foot, thus permitting of its being jointed to the rest of the floor-sheeting when that was subsequently laid.

Nine sets of girders, and five pairs of screwjacks, were provided. When the jacks had been screwed up tight, wooden blocks were driven under the ends of the channels, and in this way it was possible to remove the jacks before the cement was set, if it happened that they were required for another pier. When the men had become accustomed to the work,

Figs. 4.

six or seven piers could be dealt with in one day with this outfit. All the piers were underpinned in the manner described without any cracks appearing in the overhead arches; and as these were rendered with cement, finished quite smooth, any settlement taking place could have been readily detected. Both reservoirs are quite water-tight, after having been in use about 18 months.

The work was undertaken on the advice of Mr. Oswald Brown, M. Inst. C.E., and the operations on the reservoir first dealt with were completed under the superintendence of the Author.

The Paper is accompanied by three drawings, from which the Figures in the text have been prepared; and by two photographs, which may be seen in the library of the Institution.

(Paper No. 3580.)

"A Portable Apparatus for the Analysis of Flue-Gases." By CHARLES JOSEPH WILSON.

WHEN it is desired to ascertain the composition of flue-gases with the utmost possible accuracy, and when it is not necessary that the results should be obtained at once, the proper course to pursue is undoubtedly to collect the gas-samples in suitable receptacles and to carry out the analyses in a chemical laboratory. In practice, however, it is so often advisable to have analyses of the flue-gases made during the course of a trial, that there appears to be some demand for a portable apparatus capable of giving fairly accurate results, and sufficiently easy of manipulation to admit of its use by persons other than trained chemists.

Various forms of portable gas-apparatus have been suggested and used, but the results obtained with them have not, in general, proved to be very satisfactory. They are usually difficult to keep clean, and they fail to accomplish satisfactory determinations of carbonic oxide. The latter defect is of considerable importance, because, if combustion is regulated by the percentage of carbonic acid alone, the air-supply may be reduced to a point at which the loss of heat, due to the formation of carbonic oxide, becomes considerable. It is most desirable that a gas-analysis apparatus should be filled with mercury, and not with water or a saline solution. Amongst the sources of error inherent in the latter type of apparatus may be mentioned the tendency to alter the composition of the gas by partial solution of some constituent in a large volume of liquid, and the obvious difficulty that the reagents employed for absorption, being aqueous solutions, cannot be brought in contact with the liquid in the apparatus, as they would mix. This difficulty introduces errors in measurement, and if overcome to some extent, as may be done by the use of a considerable quantity of reagents stored in vessels in the apparatus, fresh complications are introduced by the deterioration of the reagents, particularly the alkaline pyrogallic

absorbent, which should not be kept ready mixed, but should be made as required for each absorption.

The apparatus to be described was designed with the view of avoiding, as far as possible, the various defects which experience had shown to exist in the instruments commonly used; it was employed in all the trials carried out by the recent Admiralty Committee on Naval Boilers, with good results, and is now used in His Majesty's Dockyards.

The construction of the apparatus is quite simple. Referring to

Figs. 1, in which it is illustrated diagrammatically, L is a eudiometer, in which the gases are measured, connected by a capillary tube with a "laboratory-vessel" C, in which the absorptions take place. Between L and C is a glass stop-cock A, by means of which communica tion can be established either between L and C, or betw een either C or L and the point P. Both the eudiometer and the laboratoryvessel are connected, by means of india-rubber tubing, with movable