THEORY OF VESSEL-LENGTH DETERMINATION - THE PROBLEM OF NONRANDOM VESSEL ENDS

A theoretical analysis was undertaken to examine the accuracy of algorithms commonly used to compute vessel lengths from paint perfusion experiments. The double-difference (DD) algorithm assumes that all vessels have randomly distributed vessel ends along the axis of the paint-perfused stem and that vessels do not branch. When these conditions were met, the DD algorithm overestimated die frequency of short vessels and underestimated the frequency of long vessels. When these conditions were not met, negative numbers for frequencies were outputted by the DD algorithm. Two algorithms for correcting for negative numbers were examined, one used by Zimmermann and the other used by Ewers and Fisher. Neither algorithm produced the correct result, but the correction algorithm proposed by Ewers and Fisher produced more accurate results.