SYNOPSIS This paper is largely a summary of water-surface temperature comparisons by the author on a winter-time West Indies cruise of The R. M. S. Empress of Britain. An attempt was made to determine the accuracy of observational methods under a great variety of conditions, including the most trying ones likely to be experienced. Temperatures obtained nearly simultaneously (1) from a low deck with a 2 or 4 quart tin bucket by quick dips forward of the ship's main outtakes and (2) aft in the propellor wash, and (3) in the discharge from faucets attached to the condenser intake pumps, were consistent always within 0.25° F., and differed, on the average, but 0.1°F. Reliable results are evidently procurable from the stern, where “surface” observations may, perhaps, most accurately and readily be made in cold windy weather. A record from the condenser intake pipe appears truly representative of the surface temperatures under virtually all conditions. The condenser intake temperature recorded by engineers in the engine-room log of the Empress of Britain were found to average 0.5°F. above the temperatures accurately obtained in other way. This difference appears to arise from some heating of the water about the fixed thermometers in the pumps but mostly from errors of parallax in reading. The most serious deficiency in these observations is the absence of a record of the exact time when they were made. Hourly observations on the international ice patrol ships, Tampa and Modoc are apparently of the same order of accuracy as those on the Empress of Britain. In comparison with the surface water temperature obtained with a tin bucket from a lower deck at about the same time, water surface temperatures procured by the author with a canvas bucket dropped from the bridge averaged 0.5° F. too low, and those by quartermasters with the same bucket averaged 1° F. too low. These errors were the combined result of the predip temperature of the canvas bucket, evaporative cooling of the partially filled canvas bucket after leaving the sea, temperature change of the thermometer if withdrawn for reading, and several unsystematic errors, such as occasional 5 or 10 degrees misreadings. On some other ships the average depression of the recorder canvas bucket temperatures below the condenser intake values was found to be 3°F. or more. In the Gulf Stream region north and northeast of Hatteras, winter observations from four ships gave canvas bucket temperatures averaging about 5° F. lower than the condenser intake. In cold gales over the Gulf Stream, departures in a group of 24 observations from 4 ships were so large as to have a median at 7 and upper extremes of 20 to 24°F. An analysis of some observations made on the ice patrol ships show the same tendencies when the air was much colder than the sea. Thus, Lieut. Commander E. H. Smith's observations of surface temperature stood higher than the usual canvas bucket determination from the bridge by an average of 0.7° F. for cold water (10 cases) and 1.8°F. for warm (13 cases) on and about the Grand Banks. Errors are closely related to the depression of wet bulb or air temperature below the water temperature. With air temperatures no more than 3°F. below the observed water temperatures the temperatures obtained with the canvas bucket are likely to be more than 1°F. in error in but 15 to 30 per cent of the cases. Water temperatures obtained under lower air temperatures, and especially when wind velocities are high, are likely to be too low by one-third to one-half the depression of the air temperature below the observed water temperature. With due care, involving the use of dry, stiffened canvas or wooden or fiber buckets dropped from a low deck, heaved up rapidly and as quickly observed accurate temperatures are obtainable. The use of a thermograph, the thermal element of which projects into the condenser intake pipe, is recommended, however, as much the easiest method for procuring temperatures of the general surface layer accurately under all conditions of weather. Even in late spring and summer, when surface layers are warmed more than those at intake depths, the average difference between the surface and 5 meters depth (16.3 feet) has been found to average but 0.2° C. (0.36° F.). In the 66 observations on which this average was based the surface was 0.5°C. 0.9°F.) or more warmer than water at 5 meters but 12 times, and 1°C. (1.8°F.) or more warmer but three times. The greatest difference observed was 1.52°C. (2.7° F.).