118 THE ROYAL SOCIETY OF CANADA 



specimen III. was cut from a creosoted pile which had been immersed 

 in the sea at HaHfax for several years: Specimen IV. was a piece of 

 untreated pine which had been under water during an entire summer. 

 Tests were in all cases carried out on slabs Yl" x 12" x 9". Repeated 

 observations by the ice-sandwich method gave fairly consistent 

 results for each specimen. The results are tabulated below: 



IN Tidal Waters 



From an inspection of the meteorological records for the neigh- 

 bourhood of Halifax it is found that in the month of December the 

 air- temperature may fall to — 40°F., while the sea-water remains at 

 6° or 7°C. above freezing. Throughout January, February and March 

 the sea temperatures remain steadily at -1°C., the freezing point of 

 sea- water. Under these conditions no freezing and no thawing 

 would take place on the surface of a concrete structure exposed to 

 tidal rise and fall. We conclude that most of the disintegration due 

 to this cause takes place in the early winter, chiefly towards the latter 

 part of December. For purposes of calculation we will consider air- 

 temperatures of -20°C. with sea-temperatures of +5°C. 



Although the above numerical data can only be considered in the 

 light of rough approximations, they serve a useful purpose in estimating 

 the thickness of planking necessary to protect concrete from freezing 

 and thawing as already described. If be x the maximum thickness 

 of the region in which alternate freezing and thawing takes place, 

 T the time between successive high waters (12 • 5 hours), —d°C. the 

 temperature of the air helow freezing, ( — 20°C.), ^°C. the temperature 

 of the sea above freezing (-|-5°C.), it is not difficult to show that x is 

 to be calculated from the formula 



Lp/ ^ , , , , Lp^ 



X {hf+\x) j-^ + X {h[^ + ^x) 



where the remaining symbols have meanings already assigned and are 

 all known from tests of the kind already described ' except h^, the 

 surface conductivity between wet wood and water which is negligible 

 compared to ^x. Values of x corresponding to each of the specimens 

 are given in the last column of the above table. From this it will be 

 seen that the best material for the purpose is the dense, water-satur- 



