CLYDE SEA AREA. 



671 



the same direction, i.e., when more of the stem is submerged the density appears higher. 

 Either the difference in mass is thus less than was supposed, or the difference in immersed 

 volume is greater. The brass pieces were frequently weighed, and were not found to 

 change perceptibly, but my experiments indicate the presence of both sources of error. 

 Moisture adhering to the stem makes the first density appear too low (the capillary 

 attraction is alike in both cases), and the stem instead of being truly cylindrical is very 

 slightly conical. 



It is of interest to note the frequency with which differences of each value occurred 

 in the two readings. For convenience the 1440 double observations discussed are 

 divided into 760, made in 1884-85, and 680 made in 1886-87. The former group were 

 mainly determined on shore in a properly equipped laboratory with good light, and the 

 mean difference between the two readings was 7*9. The latter set were almost all 

 observed at sea or in temporary work-rooms where the light was not always satisfactory, 

 and it is not surprising to find that the mean difference in them amounted to 8 '7. The 

 prevalence of positive values (over 95 per cent.) is a sign that the difference is at least 

 largely a constant one, and is not due to errors in reading the scale. Hence for com- 

 parative purposes the salinities deduced are quite satisfactory, and the mean values do 

 not in any case differ very widely from those deduced from chlorine-determinations. 



Table XIV. — Trustworthiness of Hydrometer Headings. 





First Group. 



Second Group. 





Discrepancy of Densities. 

 (Units of Fifth Decimal place.) 







Total 

 per cent. 













Times in 760. 



Times per cent. 



Times in 680. 



Times per cent. 





Negative Values, . 



30 



3 '9 



36 



5'3 



4-6 



0, 



14 



i -g 



19 



2-8 



2-4 



1, 





22 



2-9 



10 



IS 



2-2 



2, 





36 



4-8 



21 



3'o 



3-9 



3, 





51 



6-g 



27 



4-0 



5-4 



4, 





38 



5"i 



18 



2-8 



4-0 



5, 





64 



8-4 



52 



7-6 



8-0 



6, 





79 



104 



48 



7-2 



8-8 



7, 





47 



64 



37 



5-4 



5-9 



8, 





80 



10-6 



76 



1 1 -2 



10-9 



9, 





48 



6-5 



67 



98 



8-1 



10, 





50 



6-8 



32 



47 



5-8 



11, 





45 



5'9 



73 



107 



8-3 



12, 





37 



4-9 



39 



57 



5-3 



13 to 18, 





95 



11*2 



97 



14-2 



12-7 



19 to 24, 





13 



1-8 



12 



r8 



1-8 



Over 24, 





11 



16 



16 



2'3 '. 



1-9 



Total, 



760 



1000 



680 



IOO'O 



ioo-o 



Mean discr 



spancy, . 



7-9 



8-7 



8-3 



