Feature Article crush was typically very uniform. By contrast, specimens that failed in combined mode did not exhibit a spike and the follow-on crush was somewhat less uniform than the specimens that failed in zipper mode. The most desirable force response for energy-absorbing structural members is one where peak force is quickly reached without a spike then sustained uniformly throughout follow-on displacement. In practice the initial force spike can be mitigated by beveling (triggering) the leading edge of the tube, but this was not attempted in the present work. Energy Absorption Figure 10 shows the energydisplacement data for the XM-HP75 and XM-E75 samples and Figure 11 shows energy-displacement data for E-HP75 and E-E75 samples. After approximately 10 to 15 mm of displacement for all specimens (after the force levels out) a characteristic energy absorption rate can be calculated as the slope of the energydisplacement response. Samples that absorb more energy per unit displacement are considered better energy absorbers. Table 3 summarizes the average energy absorption rate for each sample tested in this study that could be compared. All rates are calculated as the slope of the energydisplacement curve between 20 and 45 mm displacement. For samples that exhibited two failure modes, the average energy absorption rate of each mode is calculated separately and the "Mode" column indicates how many specimens comprise each mode. Because these values often represent populations with as few as two or three repetitions, a larger sample set would be needed in order to determine the statistical differences among the tested groups. Nevertheless, it is clear that tubes made with HP-glass in the bias direction outperform the E-glass equivalent in all cases. The use of XM-glass fiber in the axial direction SAMPE Journal, Volume 53, No. 2, March/April 2017 Figure 6. Force-time data for seven specimens from the XM-E75 sample, showing high-frequency ringing. Figure 7. Unfiltered force-time data and same data after being processed with a lowpass Butterworth filter for one specimen from sample XM-E75. Figure 8. Force-displacement data for XM-HP75 and XM-E75 samples. 57