The rising of the water surface and head loss at bridge piers were experimentally studied. As a result of these tests, the followings were found.<BR>The d'Aubuisson formula, that was customarily used, overestimates head loss and incorporates hydraulic discrepancy into its formula by which the head loss per pier decrease with an increment in pier density. The Yarnell formula has avoided this discrepancy by allowing for a difference in the pier's coefficient in accordance with pier density, and with the exception of a few cases, estimates the head loss adequately.<BR>To find a more suitable formula, tests were performed and the following results were obtained.<BR>1. The condition that the flow between bridge piers were subcritical had been given by the relation N_t/B-F₂ with N and ΔZ₂/D₂ being the parameters.(N: the number of piers; t: the width of a pier; B: the width of the river; F₂: the Froude number for downstreamsection of piers;ΔZ₂: the difference of river bed elevation between the pier sections and the downstream sections.) These relationships are shown in Figs. 6 (1), (2), (3), for semicircular nosetail piers.<BR>2. Pier head-loss coefficient f was defined as f=h_p/(V₃²/2g), and then, f were related with F₃ in eq.(20).(h_p: the head loss at the piers; D₃, V₃: the depth and mean velocity of the flow between the piers, respectively; g: the acceleration of. gravity, F₃= V₃/√gD₃) In general, the flow conditions at the downstream sections are given when we seek the head loss of the piers, therefore, we need to fi nd the relation F₂-F₃ for f. The relation F₂-F₃ was obtained by eqs.(16) and (20). The head loss for piers positioned unequally in the river sections became 1.0-1.3 times the loss compared to piers positioned equally in accordance with the density of the piers.<BR>3. The flow of piers has been represented by the momentum equation, eq.(3); and α, the coefficient of piers reaction that act toward downetream was related to F₃ as in eq.(21).<BR>4. The head loss at the triangular nosetail piers was the same as the loss at the semicircular nosetail piers, and so, the method for the semicircular nosetail piers can be adapted in this case. But the head loss at the square nosetail piers became larger than these cases, and therefore, that loss must be estimated with another relationship (Figs.13 and 14).