Altitudes of celestial objects were observed with the bubble sextant from a 950 ton ship and a 12 ton craft at sea. For moderate sea conditions individual altitudes were occasionally as much as 60 minutes of arc in error and averages of 5 consecutive observations were usually correct within 30 minutes of arc. On the 950 ton ship a portion of the error was due to the vibration of the ship caused by wind, engines, etc., and a larger portion was due to the horizontal acceleration arising from the roll and pitch of the ship. If the roll were simple harmonic the latter error could be partially eliminated by taking the sextant altitude at the midpoint of the roll. A small pendulum device was arranged to sound a buzzer signal at the midpoint of the roll and the sextant observations were made simultaneously with the buzzer signal. This improved the accuracy of the bubble sextant altitudes by roughly a factor of 2, so that individual observations were rarely more than 30 minutes of arc in error and averages of 5 consecutive altitudes were usually correct within 15 minutes of arc for a moderate sea.
E. O. HULBURT, "Experiments with the Bubble Sextant at Sea," J. Opt. Soc. Am. 23, 143-147 (1933)