[ Zoi ]

at all'. But with funnels less than its own height, as oftwo. and three feet, it still raised the gage considerably.

It appears from these experiments, that in most of themachines described in the preceding section, the lengths ofthe funnels and pipes are greatly disproportioned to oneanother, and consequently the water applied to disadvan-tage. Those of Dauphiny in France are particularly faultyin this respect, the funnel being scarcely three feet high,and the pipe twenty-five or twenty-six with so small aheight of water above the choak, I have never been ableto make the jet spread near to such a degree as it is said todo in. the machines of Dauphiny, without particular con-trivances for that purpose, which will be mentioned in thesequel of this paper. The Eoix machine agrees the bestwith my experiments: but as the funnels of the others areundoubtedly much too low, that of this seems to be rathertoo high. The effect appears to be the greatest, when thefunnel is about, two thirds of the length of the pipe.

Experiments of the disposition of the air holes.

In the foregoing experiments, the simplest and most ob-vious way of admitting air was chosen, by leaving a spacebetween the funnel and the pipe. The air pipes of themachines of Foix and Languedoc answer the same end, car-rying in the air above the surface of the jet of water.As the other machines have the air holes under the jet, Itried what variations would result from this circumstance,and from making the apertures at different depths underthe throat of the funnel.

Into a pipe of six feet was fitted a funnel of four feet jand six inches below the orifice of the funnel, four holes werebored round the pipe, hoping down from without inwards :eight inches lower down, I made .another row of holes;

and