By applying a vacuum to stem segments of poplar, simultaneous determinations of the flow velocity were made using a convected heat-pulse and a radioisotope (32P). A regression equation v = 7.225 x 10−2 u—4.329 x 10−5u2 (heat-pulse velocity v on radioisotope velocity u) was fitted. This gave an average ratio u/v of c. 20 over the heat-pulse velocity range 0—20 cm h−1 with incremental ratios, du/dv, of 14 and 49 for values of v of 0 and 600 cm h−1. Using Marshall's theoretical relationship between u and v, and taking into account the percentage of vessels involved in the flow, it was possible to derive the value of the ratio u/v over the range 0–20 cm h−1 for v, and thus verify the theory. Increasing values of u/v over 20 cm h−1 are attributed to lack of thermal homogencity. Attention is drawn to the necessity to distinguish between the total lumen area and the percentage of vessels involved in the flow.