Studies of tritium radiotoxicity involving chronic /sup 3/H0H exposures in mammals demonstrate in both mice and monkeys that biological effects can be measured following remarkably low levels of exposure - levels in the range of serious practical interest to radiation protection. These studies demonstrate also that deleterious effects of /sup 3/H beta radiation do not differ significantly from those of gamma radiation at high exposures. In contrast, however, at low exposures tritium is significantly more effective than gamma rays, rad for rad, by a factor approaching 3. This is important for hazard evaluation and radiation protection because knowledge concerning biological effects of chronic low-level radiation exposure has come mainly from gamma-ray data; and predictions based on gamma-ray data will underestimate tritium effects - especially at low exposures - unless the RBE is fully taken into account.

This paper describes a quantitative approach to help evaluate and respond to safeguards alarms. These alrms may be generated internally by a facility's safeguards systems or externally by individuals claiming to possess stolen Special Nuclear Material (SNM). This approach can be used to identify the most likely cause of an alarm - theft, hoax, or error - and to evaluate alternative responses to alarms. Possible responses include conducting investigations, initiating measures to recover stolen SNM, and replying to external threats. Based on the results of each alarm investigation step, the evaluation revises the likelihoods of possible causes of an alarm, and uses this information to determine the optimal sequence of further responses. The choice of an optimal sequence of responses takes into consideration the costs and benefits of successful thefts or hoaxes. These results provide an analytical basis for setting priorities and developing contingency plans for responding to safeguards alarms.

We have developed a large, multimaterial, two-dimensional Arbitrary Lagrangian-Eulerian (ALE) computer program. The special feature of an ALE mesh is that it can be either an embedded Lagrangian mesh, a fixed Eulerian mesh, or a partially embedded, partially remapped mesh. Remapping is used to remove Lagrangian mesh distortion. This general purpose program has been used for astrophysical modeling, under the guidance of James R. Wilson. The rationale behind the development of this program will be used to highlight several important issues in program design.

Construction of the new flash x-ray induction LINAC (FXR) at Lawrence Livermore National Laboratory has been completed. Initial tuning of the machine has produced stable current pulses in excess of 2 kA at the design energy of 20 MeV, with an 80 ns FWHM pulse width, producing single-pulse radiation doses near 500 Roentgen at one meter from the target. The electronic spot size on the bremsstrahlung target is estimated at 3 to 5 mm. In this paper we will discuss the basic FXR design; running-in and tuning of the machine; emittance measurements; beam stability; switch gap synchronization; and measurements of the radiation dose and angular distribution.