a reliable steel tape properly handled, the extreme error in a number of measurements of the same line should be less than one quarter of an inch in one thousand feet. This would make the probable error of the average length considerably less than that amount. If any measurement show a greater variation from the average than one quarter of an inch to the thousand feet, it should be rejected, and another measurement should be made to replace it. This presupposes comparatively level ground for the base-line; hence, if the ground be very irregular, a greater variation may be allowed. It should, however, in no case exceed one-half inch per thousand feet.

The tape line used should be a new one for each structure, and it should be tested at the bridge shops in comparison with their standard. As a matter of precaution, it is well to test it in the field with another tape that is to be set aside as a reserve and not used unless an accident happen to the primary tape.
      For very long and important bridges, especially cantilevers with long spans, it would be well to have the tape tested by the Bureau of Weights and Measures at Washington, D. C., or by some other testing bureau of recognized standing-such, for instance, as that of the Washington University at St. Louis, Mo. The charge for such testing is usually merely nominal.

As the coefficient of expansion is not the same for all tapes, it might be advisable for extremely accurate work to have the coefficient determined for the tape to be used; but in most cases of long-span bridges this would be an unnecessary refinement.

A fifty-foot tape is long enough, and is in many respects preferable to those of greater length. The author has no use for extremely long tapes to measure distances directly between pier centres either during sinking or after the piers are finished, because this method of measurement is by no means as accurate as that of intersecting three lines on each pier and using two independently measured base-lines. The only direct measurement that is of any real value, and which can be obtained before the falsework is up, is one made on the ice. In such a measurement  care  must  be  taken  not  to  let  the  tape  touch  the  ice,   but  to  rest  it  on   plugs  driven   on  perfect  line  into  holes