This paper proposes a statistical methodology for certifying the longevity of individual lasers to be used in an undersea communications cable. In general, the strategy is to extrapolate the slope of a degradation parameter such as i (the current necessary to maintain a given level of optical-output power) when the laser is subjected to elevated temperatures. This extrapolation is made from time ii (when the test terminates) to time t p (the system life, as contracted in time by an Arrhenius relationship). An important part of the process is the calculation of the variability of the estimated current value at time t p , used to estimate the probability that it will exceed a critical failure level if·, lasers that exceed a predetermined probability will be rejected. The methodology pre sented here must, in the absence of actual data on lasers to be used in an undersea communications cable, be tentative with respect to details; modifi cations may well be necessary to accommodate idiosyncrasies of the manufac turing process and the testing equipment.

About 200 fires in Bell System buildings and adjacent grounds (excluding Western Electric) are reported to AT&T each year; the actual number of fires that occur may be somewhat higher. The dollar damage of reported fires (excluding only the $60 million fire in New York City on February 27, 1975) is reasonably modeled above the median by a log normal probability density function. This paper introduces a detailed taxonomy of fires, showing substantial differences in their frequency and costliness. The paper concludes with various special topics: (i) an analysis of employee injuries and service interruptions caused by fire; (ii) the correlation of business hours with fire frequency and building occupancy with fire severity; (iii) the methods used to fight fires; (iv) an analysis of multiple fires in buildings and of a cluster of fires in the Greater New York area in March 1975.