heel or ear. Subsequently, venipuncture specimens anticoagulated with heparin were used. Since 1973, Na EDTA has been used as the anticoagulant. Duplicate pipettes were filled for- leukocyte and platelet counts and diluted with 3% acetic acid or 1% ammonium oxalate, respectively, and each was then dispensed into a single hemocytometer chamber after 10-minute rotations and counted by phase microscopy after settling 10 minutes. Hematocrits were measured in heparinized capillary tubes. One-hundred-cell differential counts were performed on Wright's stained blood smear and used to calculate absolute concentrations of leukocyte subsets. Hemoglobin concentration after conversion to cyanmethemoglabin was measured with a Lumitron colorimeter. Hemocytometrics were used for leukocyte counts until the introduction of a Coulter counter during the third annual survey (1957). Erythrocyte counts were also performed with the Coulter instrument, and modifications in orifice diameter allowed electronic counts to replace phase microscopy for platelets after the fourteenth annual survey. The Coulter counters were superseded in 1973 by instruments from General Science for both leukocyte (MK-3) and platelet (MK-4) enumeration. In 1978, a J.T. Baker Instruments (Milford, CT) MK-4/HC platelet counting system was adopted, along with a Baker MK-40 fourparameter cell counter to measure leukocytes, erythrocytes, hemoglobin, and hematocrit. Initially, hematologic measurements in exposed individuals were repeated in duplicate or triplicate at weekly intervals. By 1962, a repeat determination was made only if an abnormal value was observed. C. Results and Discussion Specific mean hematologic values over the last five-year survey period are presented in Figures 2 to 4, segregated on the basis of exposure group and sex. It should be emphasized that the values for the Ailingnae group represent means of only 4 to 8 determinations (Table 1), and this clearly contributes to their variability. No statistically significant differences were demonstrable among the three groups exposed to radiation and their appropriate comparison populations for any of the mean hematologic values recorded during this period. Certain measurements showed tendencies to segregate on the basis of sex, however, regardless of exposure group. These included increased platelet counts in females (Figure 4) and the expected increases in various red cell parameters in males (Figure 3). Variation between sexes in platelet concentrations has been observed consistently throughout the 26-year study pe- riod and appears to represent a true anthropometric distinction in the Marshallese, as it does in Caucasians. The sporadic very low values for individual platelet counts shown in Figure 7 most likely represent technical artifacts. Despite the lack of significant variance among group means, a number of individuals exhibited sufficient deviations in certain hematologic measurements to require followup assessments. Figures 5 to 7 record individual values segregated by exposure group, sex, and time of study between 1975 and 1979 for persons whose hemograms contained values outside the ranges 4,280 to mS ford Cc ~) 5 Oa CST 10,640/y1 leukocytes, 10.2 to 17.2 g/dl hemoglobin, and 120,000 to 372,000/y1 platelets. These ranges represent values more than one standard deviation - 18 -