From an observation of the sun's altitude and azimuth and an assumed latitude a position is computed and plotted and a line drawn on the chart through this position at right angles to the azimuth of the sun as taken from the azimuth tables and laid off from a meridian. Another method is to compute positions with two assumed latitudes and plot the two resulting positions and draw a line through them. The vessel must be somewhere on the resulting Sumner line. A good determination may be obtained by the intersection of two Sumner lines obtained from two observations of the sun with sufficient interval so that there will be a change of azimuth of as much as 30 degrees to give a fair intersection. Allowance must be made for the movement of the vessel between the two observations by drawing a line parallel to the first and at a distance equal to the distance made good. An excellent intersection may be obtained by observation of the sun, and before or after it of a star in the twilight at a different azimuth. [Illustration: COURSE TO ALLOW FOR SET, GRAPHICAL SOLUTION FIG. 38.] [Illustration: POSITION BY INTERSECTION OF SUMNER LINES FIG. 39.] [Illustration: POSITION BY COMPASS BEARINGS FIG. 40.] Even a single Sumner line, however, furnishes valuable information, as it may be combined with other sources of information to obtain an approximation to the position. The vessel must be somewhere on this line, and this gives a good check on the position by dead reckoning, or an intersection may be obtained with a line or bearing of a distant land object, or a line of soundings may be compared on the chart with the Sumner line. If an observation is taken when the observed heavenly body is bearing abeam, it is evident that the resulting Sumner line will be the direction of the course of the vessel, and this fact may be useful in shaping the course when nearing the land or a danger. #Dead reckoning.# When impossible to obtain the position by any other means, it is computed or plotted from the last determined position, using the courses and distances run as shown by compass and log and allowing for effect of current and wind. Because of uncertainties in all these elements, positions so obtained may be from five to twenty miles in error in a two-hundred-mile run, depending of course to some extent on the speed of the vessel.