We found the P. of sea-water to be determined solely by ratio of the concentration of buffers, including and other weak acids, to the concentration of bases combined with them (excess base over strong acid).
Attempts were made at Tortugas and continued at Minneapolis to estimate the non-volatile buffers separately, and to find some reliable data in the literature, but this work has not reached a state for publication. It was found that phosphates and silicates are soluble enough, but it seems impossible to find enough silicate and phosphate in sea-water to account for all of the non-volatile buffers, even though we regard the results of analyses as being rough estimates. Borates were found in Tortugas sea-water and in the Princeton Marine Aquarium and in the Pacific Ocean by the senior author. An attempt was made to increase the non-volatile buffer by the addition of phosphates, but marine animals behaved abnormally in this water. The same was attempted with silicates, and plants grew luxuriantly in the artificial sea-water. No more accurate data were found when the Tortugas paper was sent to press (the paper was received for publication Sep tember 14), but about 6 weeks later the paper of Henderson and Cohn appeared, in which it was stated that artificial sea-water containing boron equivalent to 0.0015 boric acid had the same buffer value as sea-water. This led us to make a series of electrometric titration, continuing those begun at Tortugas, on a series of natural and artificial sea-waters under conditions. The results are given in figure 4, and it was estimated from this and tonometer experiments that Tortugas sea-water has a non-volatile buffer value much less than 0.001 m boric acid, whereas by the growth of diatoms and other marine Protista in a pyrex flask the same water was reduced to less than this; and Princeton aquarium water, kindly sent by Dr. L. R. Cary, had a still lower non-volatile buffer value. It is not possible to determine the exact boric-acid equivalent from figure 4, and in estimating that of Tortugas water the P. of artificial sea-waters and Tortugas sea-water at the same tension were found to be about equal.
We have not been able to obtain a sample of sea-water with as high non-volatile buffer value as artificial sea-water of 0.0015 m boric-acid content. Veatch reported evidence that the shore-water off the south ern California coast is in communication with unknown borax deposits. Dr. William E. Ritter kindly sent us a sample of this water, and it was found to give the same qualitative test for boric acid as any of the other sea-waters examined. It had a non-volatile buffer value equiv alent to Tortugas sea-water. We feel justified, therefore, in the assumption that the non-volatile buffer value of sea-water of fairly normal salinity is a more or less constant quantity and is markedly changed only when organisms are kept a long time in a relatively small quantity of sea-water. The water at the surface of the ocean, where most of the organisms live, is constantly being renewed by vertical ocean-currents.
On the assumption that the weak acids in the sea have a somewhat comparable buffer value, the error that might arise in calculating the content from the P. could be large only in case variation in non-volatile buffer is large. But is more than 100 times as strong an acid as boric acid, the dissociation constant of the former being about and of the latter may net;21,1z64 times as much NaOH as boric acid is able to, the formulae g and Na213407 (although the solution is not neutral i either case, being alkaline, due to hydrolysis). We may assum therefore, that the maximum error in estimating total fro „, due to variation in the non-volatile buffer, is negligible (not i cling any other sources of error).
The estimation of the excess base is llows : Owing to the presence of boric and traces of phosphoric a , phenolphthalein can not be used, since we have found that it es a result about 5 to 10 per ' cent too low. Methyl red is not as Give and may give too high a result, and it is better to use the same indicator we used if comparative results are desired. 100 c.c. of sea-water placed in an Erlenmeyer flask of resistance glass with etched .4 k at 100 c.c. and enough dibrom-o-cresolsulfophthalein solutio color it a distinct purple; then enough 0.01 n HC1 is run in fro a burette to turn it yellow. The flask, together with a resistance-g = or quartz beaker of distilled water, is boiled, preferably on an el 'c hot plate (alcohol flame is better than gas if electricity is not available). The sea-water will become purple again, and more acid should be added to turn it yellow and the boiling continued. Wooden tongs or a wooden model of a nut-cracker are preferable to a wire test-tube holder for handling the hot flask. The end-point is reached when just enough acid has been added to turn it yellow, and it does not turn purple on further boiling for 5 minutes, while the volume is kept approximately constant by additions of the boiling distilled water. Distilled water that has remained a long time in soft glass should not be used, and if water free from solid residue can not be obtained, it is better not to restore the volume, but to repeat the titration with the addition of all but 1 c.c. of acid before boiling and thus accelerate the evolution of CO,. Prolonged boiling after the end point has been reached is to be avoided, owing to the slight solubility of even the best resistance glass. It is theoretically possible to boil off traces of HC1 after the end-point has been reached, but this process is necessarily extremely slow at so low an acidity. Normal sea-water will require a little more or less than 25 c.c. of 0.01 m HC1 per 100 c.c. or 25 c.c. of 0.1 n HC1 per liter, and this number (for example 25) is used to denote the excess base over strong acid (i. e., the base combined with the buffer acids).