The effects of the velocity gradients on hot wire anemometry with an X hot wire probe were theoretically analyzed. Analysis showed a strong possibility that the values obtained from the traditional way of measurements with an X hot wire probe involve serious experimental errors caused by the velocity gradients. A new method of measuring which eliminates those errors was proposed based on the results of the analysis. The new method includes the use of two X hot wire probes, one of which is the reflected image of the other, and the taking of arithmetic means of the values obtained from those probes. The validity of the analysis was confirmed by measuring a turbulent flow in a square duct with rough walls on two opposite sides. The results of measurements showed that the new method is very effective for the measurement of quantities, such as the secondary flow velocities and the turbulent shear stresses, quantities which are very sensitive to the effects of velocity gradients.