Immediately after treadmill walking ended, we measured heart rate, blood pressure (n = 9), Tre, skin temperatures, and relative humidity inside the uniform, and participants rated their perceptions. We also drew another 7-mL blood sample and analyzed it in the same manner used for analysis of the pre-exercise sample. Plasma volume change from pre-exercise to postexercise was calculated using the method of Dill and Costill.20

Next, participants left the environmental chamber, and after removing their equipment and uniforms, body mass was recorded. The postexercise body mass was subtracted from the pre-exercise value to determine mean whole-body sweat rate, corrected when necessary for urine and fecal excretion.

Finally, participants sat quietly and drank water ad libitum. During recovery, Tre was monitored to ensure that thermoregulation returned to normal. Before leaving the Human Performance Laboratory, participants were advised to drink at least 150% of the water weight that they had lost, preferably combined with a meal, to encourage their return to a euhydrated state during the next 24 hours. Within 14 days of completing experimental trials, dual-energy x-ray absorptiometry (DEXA) (Prodigy; Lunar Corporation, Madison, WI) was used to determine body fat percentage, total fat mass, fat-free mass, and lean body mass.

An a priori power analysis was performed to determine the adequacy of sample size. Because the statistical variance of each outcome variable was different, it was impossible to justify sample size on the basis of a power analysis of all variables. Thus, we elected to use Tre as the critical variable because of its importance in participant safety. A minimum sample of 8 participants was identified for a 2-tailed statistical test with an ? level of .05 and with a desired minimum power of 0.8. In consideration of this calculation, we conservatively enrolled 10 participants.

Treatment effects were evaluated using a randomized, crossover design. All data were statistically analyzed using a 2-way repeated-measures (uniform × time) analysis of variance (ANOVA) and are reported as mean ± SD. When we found differences, we used the Newman-Keuls test for post hoc analysis. The magnitude of observed effects (ie, effect size) was calculated via the formula d = (mean1 ? mean2/?), where mean1 refers to the mean of the experimental group, mean2 is the mean of the control group (or second experimental group), and ? refers to the pooled SD (ie, the average of the SDs of the 2 groups). Bonferroni corrections with post hoc t tests were used to determine pairwise differences among uniform type and time. A 2-tailed, paired t test was used to evaluate differences between the means of variables measured only pre-exercise and postexercise.

To evaluate relationships between key outcome variables and each participant's personal characteristics, linear regression analysis was performed to analyze 2 different relationships: Tre increase versus lean body mass (kg) and treadmill exercise time versus total fat mass (kg). Other variables, including skin surface area (m2), age, height, and body mass, were compared with Tre and treadmill exercise time via linear regression analysis. The ? level was set at .05. We used Statistica (version 5.5; StatSoft Inc, Tulsa, OK) to analyze the data.

The environmental conditions during all laboratory experiments were not different for air temperature (CON = 33.0 ± 0.7°C, PART = 33.1 ± 0.8°C, FULL = 33.0 ± 0.7°C) (F2,100 = 0.50, P = .68) and relative humidity (CON = 48.7% ± 5.5%, PART = 49.4% ± 5.3%, FULL = 48.0% ± 5.9%) (F2,87 = 0.28, P = .85). The entering body mass (CON = 114.87 ± 8.48 kg, PART = 114.78 ± 8.54 kg, FULL = 115.65 ± 8.47 kg) (F2,27 = 0.03, P = .97) and urine specific gravity (CON = 1.015 ± 0.008, PART = 1.018 ± 0.007, FULL = 1.016 ± 0.007) (F2,24 = 0.36, P = .70) of participants were not different across days, indicating that they entered all uniform experiments in a similar state of hydration. Food intake and activity during the 24 hours before experiments were consistent across treatments.

The relative humidity near the skin surface varied among the 3 uniform conditions. The relative humidity for the CON condition was 47% ± 4% throughout the entire experiment; the relative humidity for the PART condition ranged from 75% ± 9% to 89% ± 4%; and the relative humidity for the FULL condition ranged from 69% ± 11% to 90% ± 5%. Although the relative humidity values for the PART and FULL conditions were not different throughout the experiment (F2,22 = 0.028, P = .9), the CON condition was different from the PART and FULL conditions at all time points (F2,27 range, 20.6–223.4; P < .001).

Exercise Time and Perceptual Ratings

All participants completed at least 15 minutes of treadmill exercise. However, the CON, PART, and FULL conditions induced different physiologic strains, resulting in a different number of participants completing 60 minutes of treadmill exercise. Seven participants completed the entire 80-minute protocol in the CON condition, 3 in the PART condition, and 1 in the FULL condition (Table 2). Investigators did not need to stop exercise for any participant, because no participant reached the predetermined Tre or heart rate safety limits; instead, physical exhaustion caused participants to prematurely terminate experiments. As an indicator of exercise-heat tolerance, Figure 1 presents the time to exhaustion for the CON, PART, and FULL conditions.