incidental to the lake, and by virtue of it, the amount of excavation and the attendant difficulties in the Culebra Cut are greatly reduced.

The lake provides 23 miles of canal channel, or nearly half the total length, and gives a width of 1000 feet for 16 miles, 800 feet for 4 miles, and 500 feet for the remaining 4 miles; the average width is nearly 900 feet, while the rest of the canal averages less than 450 feet. Not only in width, but also in depth the lake channel offers an advantage, for while the rest of the channel is limited to an ample depth of 45 feet, the lake offers a maximum depth of about 75 feet, and is nowhere less than 45 feet along the navigable channel. These generous dimensions will facilitate navigation and will allow vessels to approach their ocean speed.

Besides being such a valuable asset to navigation, Gatun Lake solves one of the most difficult and most vital of all the problems involved in the canal construction. We are familiar with the characteristics of the Chagres River. This wild and variable stream is immediately tamed and calmed on entering Gatun Lake. Its waters, which form and replenish the lake, may be likened to a beast of burden quietly carrying the ships to and fro, supplying the lifting force that passes them through the locks, and the power to drive the generators which light the canal, operate the machinery, and which may, later, operate the railroad.

While great ideas and great accomplishments may be briefly abstracted in picturesque terms, the knowledge so given is superficial if unaccompanied by a more intimate consideration of the principles involved,

and of the studies and investigations which attended them. Nothing may be left to surmise or conjecture, no assumptions may be made, unsupported by masses of the best evidence available. Where the problems are new and no direct evidence can be obtained, the best engineering judgment, based on experience, must be brought into play.

An investigation had first to be made as to the sufficiency of the water supply. The lake, once it is formed, will suffer losses from at least five different sources: 1st, evaporation; 2d, seepage, or groundflow; 3d, leakage through the lock gates and spillway gates; 4th, water required to pass ships through the locks, and 5th, water to develop power, if a sufficient amount remains available.

Evaporation depends on the wind and the hygrometric state of the air, and also on the area of the lake. At the normal elevation of 85 feet above sea level, the area of the lake is 163 square miles. For certain reasons that will be discussed later, the elevation of the lake may, when actually placed in service, vary from 80 to 90 feet above sea level, and the area of the lake will vary correspondingly from 153 to 173 square miles. Evaporation continues from day to day, and, unfortunately, is the greatest when rainfall is the least. The length of the dry season is, therefore, of importance. To provide for the driest future year, the weather records as far back as available are studied, and the driest year taken as a standard, with an allowance for even more unfavorable conditions. Fortunately, the French under the New Company, differing from the

de Lesseps Company, made continuous and careful observations of all meteorological and hydrological features of value. The Americans have continued these observations with great care and completeness. Evaporation pans have also been exposed to secure direct evidence which would bear some relation to the rate of evaporation from the lake. From the best evidence available at the time, the probable rate of evaporation was found to be about one-fourth of an inch per 24 hours. This has been computed to equal a loss of 930 cubic feet per second, continuing during the dry season. Later observations indicate one-sixth of an inch daily during the dry season, thus reducing the computed probable loss.

The loss by seepage is dependent on the character of the soil forming the bottom of the lake, and upon the head or pressure of water at any particular point. To clearly understand its character, we may note that an ordinary river in reality includes more than the flowing water which is visible between its banks, in that the ground along the river contains water which to the eye seems quiescent, but which actually has a flow, extremely slow, but always moving toward the river and down the valley with the river. Its rate of flow depends on the character of the material, the frictional resistance, and the distance to be traveled; it is comparatively rapid in sand or gravel, and is reduced to a minimum in clays and rocks. The seepage from Gatun Lake will be of an allied nature, and it remains to estimate the amount. The engineers made careful studies of the bottom of the lake by borings,

test pits, and geological surveys. Specially careful examinations were made at those points where the ridges between the lake and the adjoining valleys are narrow and low. It was perfectly possible that gravel strata or porous coral deposits might exist which, communicating with the sea, might discharge the waters of the lake as through a sieve. The engineers satisfied themselves that no such condition existed, and their judgment was confirmed by a Board of Consulting Engineers appointed in 1908 by President Roosevelt. The probable seepage was estimated to be 85 cubic feet per second, or less than one-tenth the rate of evaporation during the dry season.

The loss of water through leaks and imperfect seatings in the many valves and miter-gates of the locks, and the 14 gates of the spillway, depends on the accuracy with which the devices are made and the care used in the maintenance. The commission followed correct principles in using the utmost care in designing and constructing them, and yet assuming a rather heavy loss of water from incomplete closure or accident. The amount lost is estimated to be 275 cubic feet per second, the equivalent of 500 ordinary city fire streams.

The amount of water found necessary for developing electric current for lighting the canal, and operating all the machinery is estimated at 275 cubic feet per second, based on the required amount of current and the efficiency of the apparatus.

The amount of water required for lockages is dependent on the design of the locks, the amount of traffic, and the size of the vessels, for the locks are so