deviation of calibration response values is about 7%. Figure B-5-3 and Table B-5-3 show the response data from 27 February onward. Evidently, a decrease in response of the detector occurred between 17 March and 21 March. Between these dates the detector was removed from its barrel, another tried and found unsuitable, and 496 reinstalled. The mean response from 25 February to 17 March was 5979 + 31; mean response from 21 March to 12 July was 524 + 20. The ratio is 1.11. Statistical analysis of the two sample populations (27 February to 17 March vs. 21 March to 12 July) was conducted using the Student's "t" technique (conducted by J. J. Giacomini of Desert Research Institute). Comparison of the difference between means of the two populations with the standard deviation of the differences gives a "t" value whose magnitude implies a difference in the two populations. The probability of observing this large a "t" value for the null hypothesis, i.e., that the two sample populations are not different, is less than 0.001. A similar examination for the 137cCs and 80¢9 peaks gives the same conclusion. Table B-5-4 gives a summary of the basic statistics. There are three corroborating data points: (1) Detector effective area measurements by EG&G at Las Vegas before island use show ratio of 1.12 for detectors 393/496. Measurements on 15 and 22 July at Ursula give a ratio of 1.22. (2) Calibrations performed in May 1978 for the soil sample screening method give a ratio of 1.19 for detector 393/496. (Recall that the March 1978 field experiment gave a ratio of 1.10 for these two detectors.) (3) Efficiency measurements at the ERSP Enewetak counting laboratory for detector 496 show a ratio of 1.16 for 241 4m, comparing 2 February to 25 July data. Recommendation It is recommended that detector 496 be corrected by multiplying its readings by a factor of 1.16 for degradation during the period 21 March to 12 July. This is based on the field calibration data averages, the counting laboratory results, and a comparison of detector effective area as measured at Ursula on 15 July, with the effective area of 19 used in the IMP calculation program. The factor of 1.10 to account for the smaller active area of 496 relative to the other detectors is still applicable for the period 25 February to 12 July. The correetion factor recommended for 21 March to 12 July data is 1.10 X 1.16 =1.276 =1.28. Sam RESPONSE — MORNING + NOON * AFTEANOON 500 + BE Bs a 105 + + + 126 JULIAN DATE, 1978 145 + o “ . 165 — 186 FIGURE B-5-3. AMERICIUM 241 CALIBRATION RESPONSE FOR DETECTOR 496 27 FEB TO 12 JULY 1978 B-5-4