for typhoid deaths are set back one month to bring out any relationship existing between them and the bacterial pollution. Even making due allowance for the expected fall increase in typhoid fever from other causes than water supply there is agreement between these curves which is more than mere coincidence. Further discussion of the importance of the seasonal variation here disclosed will be deferred until the matter of interpretation and standards is considered. The ratios of Table I, however, are so definitely established that they will be employed in the further discussions of the extent of pollution of the boundary

waters.

EXTENT OF POLLUTION.

Taking up now the actual extent of pollution shown in the Progress Report, this may be briefly summarized as follows, the pollution being expressed in each case in terms of B. coli per 100 cubic centimeters.

ABOVE LAKE HURON.

The waters of Rainy Lake showed a minor but quite unexpected pollution. The river itself shows The river itself shows the effect of pollution from Port Frances and International Falls, which, in July, 1913, resulted in an average B. coli value across the river below these cities of 270. This pollution remains practically constant throughout the length of the river and distinctly affects the waters of the Lake of the Woods in the vicinity of the river mouth.

Lake Superior showed extensive pollution in the Thunder Bay region, especially in front of Port Arthur and Port William.

The St. Marys River contained 17 B. coli per 100 cubic centimeters at its source, increasing to 281 below the "Soo" and continuing at this figure into Lake Huron. The Straits of Mackinac were found to be quite pure.

ST. CLAIR RIVER AND LAKE.

The lower end of Lake Huron showed unmistakable evidence of steamboat pollution.

The St. Clair River was found to be reasonably pure at its upper end, steamboat pollution accounting principally for the 9 B. coli found. Above Sarnia the figures reached 41 during August, 1913; and below Sarnia and Port Huron they ranged from 22 to 763 at various cross sections. Examinations at single stations during the first 15 days of August gave results of from 106 to 1,228. The average of all examinations of this stream made during the season, 1,198 samples, is 254 B. coli per 100 cc.

The mouth of the Black River gave an average value of 3,400, showing its gross pollution.

Lake St. Clair showed very little effect from the pollution of the St. Clair River. It was found to be moderately polluted by shipping, especially at its lower end.

DETROIT RIVER.

The Detroit River was more fully studied than any of the waters hitherto described. The discussion of seasonal variation is based largely upon the data of this river. Whereas the average condition of the streams that have been dealt with is given with sufficient accuracy by averaging the results of cross sectional studies, such a procedure is not satisfactory in such a highly polluted stream as the Detroit. The influence of the highly concentrated shore pollution can not be properly weighted in any selection of sampling stations however carefully made.

On the other hand, the average value of a large number of samples will tend to eliminate sampling errors in any one section of the river, and a check upon the general accuracy of the average is obtained by a comparison of results at various cross sections between which little or no additional pollution is contributed. The data on the Detroit River are sufficiently extensive for this purpose but extend over varying seasonal ranges and because of the marked seasonal changes already described do not permit direct averaging.

Table III contains a summary of all data on cross sections below the River Rouge in the region where additional pollution is small in contrast with that added above. In order to determine the general average character of the water in this region the data have been submitted to the following treatment.

Individual series of results on one cross section are first averaged to give an approximate cross sectional value at the mid date of the series. This value is reduced to a mean seasonal value by use of a seasonal factor derived from the seasonal distribution curve, Plate I. The two or more series of determinations made at each section and reduced to mean seasonal values are then averaged, each weighted by the number of samples involved. The result is put down as the average cross sectional value at this point during the season May-September. Four sections are thus treated, the first immediately below the River Rouge, the second at the upper end of Grosse Isle, the third opposite Grosse Isle, on both sides, and the fourth below Grosse Isle.

The average values obtained at these four sections by this procedure are in substantial agreement and indicate the extent of pollution of the Detroit by the population upon its shores. This agreement also indicates that the methods employed in this analysis and the seasonal distribution results derived from the data of four rivers are sufficiently accurate for the present purpose. The weighted average of the four sections. is 8,030 B. coli per 100 cubic centimeters, a figure which will be taken as representative of the average extent of pollution May 1 to September 30, 1913.

From this mean value and the curve of seasonal variation there can now be derived a corrected statement of conditions during the remainder of the season covered by the data. This result is shown in Table IV.

TABLE IV.-Extent and seasonal distribution of pollution, Detroit River.

[Computed from mean seasonal value and seasonal distribution curve.]

ern end of the lake showed some pollution from navigation and some shore pollution confined quite closely to the immediate shore waters.

The waters of the Niagara River were studied extensively, although not in such detail or over as large seasonal range as in the case of the Detroit. The peculiar nature of the channels between Buffalo and the Falls makes it difficult to obtain, from the data, cross-sectional values as a measure of the total pollution. Four series of examinations are, however, available for the purpose, two near the mouth of the river, one a single set of 40 samples just below the Falls, and one near the lower end of Goat Island. These results have been treated as in the case of the Detroit River, in Table V.

1.29

10,350 2.31 18,550

53 4,260

July..

August..

September...

1.00

8,030

Mean...

Month.

B. coli Seasonal per 100 factor. cubic cen

timeters.

This table gives a reasonably accurate picture of the actual extent of the pollution of the Detroit River as affected by seasonal variations. Other causes of variation and their effects will be discussed in a later section.

LAKE ERIE AND NIAGARA RIVER.

The western end of Lake Erie was found to be so noticeably affected by the pollution entering by the Detroit River as to be unsafe for use on steamers. Outside of a line from Point Pelee to Sandusky, however, the water was found to be quite pure. The east

It is now possible also to compare the mean values obtained upon the Detroit and Niagara Rivers with the respective contributing populations, the stream volume being practically the same in the two cases. This comparison is given in Table VII.

LAKE ONTARIO AND THE ST. LAWRENCE.

The pollution of the Niagara River was found to extend into Lake Ontario for 10 or 12 miles and at times 16 miles from the river's mouth. The direction and extent of this influence was found to be condi

The lower end of Lake Ontario was found to be almost sterile before the season of navigation. It was slightly polluted later in the season by boats. The Kingston water front was found seriously polluted. Minor and intermittent pollution was observed about the Thousand Islands, rendering the waters unsafe for local domestic consumption at these points.

In the region about and below Ogdensburg an average B. coli value of 20 was obtained May 9-16, and of 138 August 1-27. Reduced by the seasonal factors, these become 59 and 41, respectively, average 50 for the May-September season. The number of examinations is too few for definite comparison with the other rivers, but taking the mean discharge value of 240,000 second-feet for the St. Lawrence as compared with 210,000 second-feet for the Niagara (Lake Survey Bulletin No. 24, 1915, U. S. War Department), the pollution at this point is representative of a population of about 6,000 on the basis of the Niagara and Detroit values. It is not known what proportion of the 21,000 total population at this point is actually connected with the sewerage system.

THE ST. JOHNS.

The St. Johns River showed pollution amounting to about 125 B. coli per 100 cubic centimeters in the late fall, October 7-November 4, which in the light of the seasonal curves obtained elsewhere, indicate a mean May-September pollution value of at least 9,400. This is excessive in comparison with the Detroit and Niagara River figures, population and stream flow considered. Assuming a low run-off of 4,000 secondfeet for the St. Johns in October, the B. coli figures per capita are about 65 times as great in this case as in the others. This abnormal condition was attributed in large measure to the potato-starch factories located along this river, the waste from which contains large numbers of bacteria. In this case, therefore, the B. coli do not represent pollution of a dangerous sort.

VARIATION FROM AVERAGE CONDITIONS.

While the average results derived in this way are convenient and necessary in discussing the extent of pollution and remedies therefor, it is particularly important to note the departures from the average, both seasonal and as to cross sections. The long time seasonal variations have been sufficiently set forth in Tables IV and VI. In addition to these there are short-time variations extending over periods of from one to seven days, due to actual fluctuations in stream discharge. Maximum fluctuations in this factor may be taken at approximately 25 per cent either way of the mean, and will result in departures of a similar magnitude from the mean index of pollution. Geographically, or as regards the cross section, there is even wider departure from the average conditions, the maximum values ranging from 2 to 15 times the mean cross sectional value. These variations are of immediate interest in connection with water supplies and indicate the importance of a proper study of the location of waterworks intakes.

SIGNIFICANCE OF THE POLLUTION DATA. The significance of the bacterial evidence of pollution must be considered from the point of view of its magnitude, its seasonal and geographical distribution, and its actual relation to the bacteria of disease. The practical limitations of water purification must also be considered. The first two of the matters have already been discussed.

RELATION TO WATER SUPPLIES.

Experience has shown that there is a distinct relation between B. coli values in a polluted water and the incidence of typhoid fever in communities using such water. This relation is not well fixed numerically, but with improved methods of study and with the general reduction in typhoid fever rates, it is becoming increasingly apparent that high drinking water standards are essential. Where formerly a standard of 50 B. coli per 100 cubic centimeters was considered permissible in a drinking water, the tendency of the times, based upon experience, is toward more severe standards, such as 10 or less per 100 cubic centimeters. With improved methods and decreased cost of water purification it is not improbable that even better results will be demanded in the future, and will be found to be justified.

It is obvious, therefore, that the waters of the connecting rivers, which are most important in connection with municipal water supplies, are not fit for domestic consumption without purification, and could not by any feasible method be protected to that extent.

OPINION OF CONSULTING ENGINEERS.

This matter was discussed before the Commission by a board of consulting engineers, and the following