After the base-lines are measured and the hubs are put in, the next step to take is to measure the six principal angles of the triangulation. These should be measured with the greatest accuracy continuously around the limb of the transit, making from ten to thirty readings of each angle, according to the degree of refinement required. The instrument should be graduated for accurate work as fine as twenty seconds, or preferably ten seconds. A heavy transit with a good, solid tripod will usually give better results than those obtained by using a lighter instrument. The sun should never be permitted to shine on the instrument when the angles are being observed, as it is impossible to make accurate measurements under such a condition.

In keeping notes of triangulation-work a record should be made of the date, the temperature, the condition of the weather, the direction and approximate velocity of the wind, and the names of the transitman and picketman.

If long sights are to be taken, the picketman should be provided with a pair of field-glasses to enable him to see the transitman's signals; otherwise much time and labor may be spent to no purpose. Long sights should never be taken towards the sun when it can be avoided.

The error of all three angles in each of the two main triangles should not exceed two seconds in important work. Of course it is not necessary to go to any such refinement in short-span bridges; but in very long ones the error might well be reduced as low as one second. If the error in a triangle be found too large, it may be possible to avoid measuring all three angles again by looking over the notes and ascertaining from the weather conditions which angle is most likely to be at fault, then measuring this angle anew. If the second average angle reduces the total error in the triangle to within a proper limit, all right; but if not, the other two angles will also have to be measured a second time.

On the same principle, if, in a group of measurements of one angle, one or two readings be found to differ greatly from the others, they may be thrown out when obtaining the average.