It is a moot question whether, in making an earthwork embankment, dependence, as far as stanchness is concerned, should be placed upon the puddle wall alone or upon the embankments on either side, and especially upon the up-stream side in addition. Supposing the former idea prevails, then it can be of little moment as to how or of what material the bank on either side is made up--whether of earth or stone--placed in thin layers or tipped in banks of 3 ft. or 4 ft. high; but the opinion of the majority of engineers seems to be in favor of making the banks act not merely as buttresses to the puddle wall, and throwing the whole onus, as it may be termed, of stanchness upon that, but also sharing the responsibility and lessening the chances of rupture thereby. But to insure this, the material must be of the very best description for the purpose. Stones, if allowed at all--and in the author's opinion they should not be--should be small, few, and far between. Let those that are sifted out be thrown into the tail of the down stream slope. They will do no harm there, but the layers of earth must not approach 3 ft. in thickness nor 1 ft.--the maximum should be six in., and this applies also to the puddle. Let the soil be brought on by say one-horse carts, spread in six inch layers, and well watered. The traffic of the carts will consolidate it, and in places where carts cannot traverse it should be punned. In the Parvy reservoir dam a roller was employed for this purpose. It comprised a small lorry body holding about a yard and a half of stone, with two axles, on each of which was keyed a row of five or six wheels. At the Oued Meurad dam, in Algeria, 95 ft. high, constructed about 23 years ago, the earthwork layers were deposited normal to the outer slope, and as the bank was carried up the water was admitted and allowed to rise to near the temporary crest, and as soon as the bank had settled, the earthwork continued another grade, and the same process repeated.