summaryRF pretreatment of lumber does not afYect strength. 0The amount of pinene lost into the headspace during low-VOC RF-treatment of wood approximately corresponds to the amount of material lost from the wood. Virtually all the pinene can be removed from the low-VOC reactor with steam, suggesting that pinene can be collected when the small amount of steam released during low-headspace treatment is condensed. Temperature and moisture loss profiles for particle at 105OC has been modeled using experimental data It 13OoC and 16OoC. The VOC-temperature curve fiom dried particle shows a break at about 156"C, the boiling point of a-pinene, demonstrating that pinene boil-off occurs beyond this threshold. VOC release from dry particle has been successhlly modeled. The transport of VOC from sapwood to the atmosphere for pine is faster than the corresponding movement from heartwood to sapwood. Seasonal variations in pine extractives are small. Effect of RF-irradiaticm on strength (MSUBPST)Twenty five Southern pine 2" x 6" x 8' boards were machined into 2" x 4" pieces. Next, the 8' boards were cut in half into matched pairs. One of the two was irradiated with RF, while the other served as a control. Both sets were dried under a conventional temperature-time based schedule. The lumber was removed from the kiln when it reached an average dry-basis MC of 15%. After drying, the moisture was equalized to about 12%. Two boards from each charge warped during drying. The lumber was machined into test specimens, and tests were performed as per ASTM Section D-143 and Appendix Q. Temperature and other changes are recorded in Table 1. Tables 2 and 3 compare strengths of the control and irradiated boards, respectively. Tensile measurements were made on two samples cut fiom each board; the data are separated into stronger and weaker groups respectively. Paired T-tests (single-tail) with a=0.5 showed minor differences in shear, MOE, and MOR. A test of hypotheses and a test through confidence intervals also showed that the null hypotheses (Ho:u=uo) for both MOE and MOR were rejected; i.e. the control and sample were significantly different. However, range and a sample mean control charts showed a possibility of a Type I error (€30 rejected when Ho is true), since the process was on both sides of the process 1