G.Lapk. The 2011 for bubble sextant mentioned 53´ presetting is for refraction + 1/2 diameter : refr. 37´ ; diam. 32´. By refraction the sun shows an apparant elevation which is greater than the true elevation. Also see figure 3 illustration , p.27 of EJN 2008. Since a bubble sextant needs no correction for "dip" , the artificial equator being parallel with the celestial equator , correction is for refraction (plus parallax for the moon) only.
For the marine sextant things are different. The marine sextant registers on the visible horizon which latter ´dips´ lower when observed from altitudes above sea level. Depending on what nav table is used , somewhat different figures are found , the difference coming from rounding as well of development of empiric magnitudes , and of temperature/pression corrections applied yes or no . 25´.2 and 33´.0 in 2008-2011 issues are for dip only (not refrac etc).
The general formula for dip is : Dp = sq.rt 2 H/R with H = altitude in meters , R = radius of earth. The outcome is in radians and should thence be multiplied by 57.296 to obtain dip in arcminutes.
Example : Altitude over sea 1,000 ft = 305 m . Dp = sq.rt. 2 x 305 / 6,400,000 m = 0´.009760 x 57.2296 = 0 deg 33´34" , or 33´ finished. For this same case H.O.no.208 says 31´ . In navigation , finishing is subtractive.
If we follow H.O.208 the marine sextant setting for a sunrise observation @ 1,000 ft should be : I. set the index screw to (+) 31´ by which II. the horizon will show up ahead with the instrument held horizontally. III . Apply green filter ´dark´ and wait until U.L. of sun just clears the horizon. IV . Note the time point in GMT for U.L. & horizon tangency. V . Consult your precompted sunrise-time-coordinates listing , the figures closest to observation time are the right ones. VI. Accuracy : 6 miles or better uncertainty for experienced navigator.