The present study was aimed at investigating patterns of development of length measuring behavior. Seventy-one 1st-3rd graders were given 10 measurement tasks individually. Each task required S to compare the length of 2 straight lines. Four of the tasks were concerned with formation of schemata of measurement, i.e., the ability of S to judge the relative length correctly by proper measuring procedures, which he applied spontaneously or in imitation of E's demonstration. The other 6 tasks were to test conceptual abilities essential to measurement, in which conditions were such as to call for logical inference on the part of S. They were concerned with conservation of length under change of position, conservation against Müller-Lyer illusion, transitivity, conservation of sub-divided whole, disregard of non-equivalent units, and mutual conversion of 2 different units.<br>Following points were suggested from results:<br>1) In the direct comparison tasks, almost all of the Ss judged the relative length of 2 lines correctly. Successfull respondents in the indirect comparison task (using intermediary items) and the unit iteration task were nearly the same in number, i.e., 2/3 of the 2nd graders and all but few exceptions of the 3rd graders. In the indirect comparison task, most errors were due to the incorrect combination of two direct comparisons which were made with the intermediary. Given a ruler as an intermediary, however, 70% of even the 1st graders responded correctly. This means that numerals assigned facilitated combination of two direct comparisons.<br>2) As for conservation and transitivity (they were observed to be highly correlated), the proportion of correct respondents reached 2/3 of the 2nd and the 3rd graders. On the contrary, 2 tasks involving operational comprehension of unit iteration were much more difficult; only 1/3 of the 3rd graders successfully applied a common unit, disregarding non-equivalent unit-marks.<br>3) Though each schema of measurement has some basic conceptions as prerequisites, a pupil without these conceptions often succeeded in applying the schema and in judging the relative length correctly, if intuitive regulation was permitted.<br>4) Some children compared the length of 2 lines correctly by counting the number of units. At the same time, they often made errors by responding only to the number, disregarding the size of unit. Measuring behavior of pupils of these ages was not clearly differentiated from numerical behavior and was highly influenced by it.