1914] Sumner, et al.: Physical Conditions in San Francisco Bay 35 
Each of the foregoing tables likewise gives the mean figures for 
the bay as a whole during the period in question, viz., the means for 
surface and bottom temperatures, those for flood and ebb, and finally 
the grand average for the period. These means for the bay as a whcle 
are not simple averages, such as would result from adding up the 
figures in the various vertical columns and dividing each sum by the 
number of stations. Instead of this, the average for each of the four 
groups of stations (see p. 22) has first been obtained. Each of these 
averages has then been multiplied by the figure representing the per- 
centage of the total water of the bay contained in the section in 
question. Finally, these products have been added and divided by 100. 
Such a “ 
mean temperature of the bay as a whole than the simple mean of all 
the station figures would have done. 
Table 9 gives the yearly average for each station and for the bay 
asa whole. The surface figure for each station was obtained as follows: 
The mean of the surface temperatures for all the periods was com- 
puted separately for the ebb and flood tides. The mean of the two 
resulting figures is that which has been employed in the table for the 
year. The same procedure was followed in computing the bottom 
figures in this table. 
The flood and ebb figures were similarly obtained, each representing 
the mean of the surface and bottom figures for the corresponding 
phase of the tide. Since the object of presenting the flood and ebb 
figures separately is to permit of a comparison between the two, these 
figures have been omitted in the case of those stations for which the 
observations at one or the other phase of the tide were not complete.** 
For the station means, the surface and bottom figures of this table 
have been averaged as in the preceding cases. In averaging the various 
station figures to obtain the grand averages for the bay as a whole, 
the same method has been employed as has already been described for 
tables 3 to 8. 
? 
weighted’? mean obviously represents more fairly the 
12 Viz., when for a given station the flood or ebb figure was lacking for one 
or more of the periods. Suppose, for example, that the flood figure, but not 
that for the ebb, were available for the season of highest temperature. The 
mean temperature for the flood observations would be unduly high, as com- 
pared with that for the ebb observations. Erroneous conclusions might readily 
be drawn. This is a type of error against which it has been constantly necessary 
to guard. 
