ship for crater depth may approach the fourth root; this conforms with the crater- prediction curves in Reference 7 (Figures 32, 33, 34, and 45). Thus, especially based on the conclusions derived in Reference 10, (made partiaily possible by the data of Castle Project 3.2) considerable increase in reliability has resulted with respect to predictions of craters produced by megaton detonations. 4.3 TREE-STAND STUDIES Operation Castle presented an opportunity to make measurements on natural tree stands several times larger than the Operation Upshot-Knothole experimental tree stand. Even though the natural stands were composed of tropical trees found at the EPG, break- age data was considered desirable, since continental tests ir forested areas were not planned. During Upshot~Knothole, an artificial stand of trees 320 feet long by 160 feet wide compos2d of 145 Ponderosa pine trees averaging 51 feet in height, had been exposed at a 4.5-psi peak static overpressure. The stand was instrumented elong its length and across its width with ground~level static-overpressure gages, as well as dynamicpressure gages at three elevations located 250 teet from the front of the stand. Groundlevel pressure measurements had showed no significant attenuation in peak static overpressure or increase in rise times. Upshot~-Knothole results had also indicated thut the prediction system for isolated trees was conservative when applied to small coniferous tree stands. However, in view of the unknown derree of att: nuation for large stands and the tenuous nature of militarydamagecriteria for trees, damage predictions for isolated trees were assumed representative for tree stands. Thus, from all available data, a general breakage-prediction system had been developed that represented various levels of breakage probabilit: for tree stands. The prediction system could be applied to idealized tree stands to determine damage by various-yield weapons, using height-of-burst curves modified to include wave form, where damage criteria were based on length of stem down per acre. Forthree general tree-stand types, isodamage curves giving light and heavy damage had been prepared for inclusion in TM 23-200 (Reference 7). Sample plots were selected on three small, naturally forested islands of Bikini Atoll-_— Uncle, Victor, and William. These islands spanned a desirable predicted-overpressure region for the expected yield from Shot 3 ranging from heavy damageto light or no damage. It was essential that a substantial portion of the trees remain intact as a group, giving a graded series of damage to correlate with the previously developed tree-breakage prediction system. In spite of the unexpected low yield of Shot 3, Project 3.3 achieved basic damage data. The unexpectedly large yield of Shot 1—blast incident from the opposite direction of Shot 3—caused heavy damageto the tree stands on William and Victor Islands and light damage to the upper portion of the stand on Uncle Island. Shot 2—-blast incident from the same direction as Shot 1-——caused no additional damage. The Shot 1-Shot 3 situation proved to be very fortunate. Because of the opposite directions of blast incidence and extreme yield difference, heavy damage from Shot 3 only extended to just beyond the light damage region of Shot 1. Thus, two sets of graded’ damage data were secured instead of one: from a high-yield device with long positive-phase duration (15.0 Mt, 2.5-psi peak static overpressure, 10-second positive-phase duration) and from a medium-yield device with shorter positive-phase duration (130 kt, 4.5-psi peak static overpressure, 1.2-second positive-phase duration). The principal tree growth on the three islands selected consisted of five main compo66