Attn Mark - New measurements to look at

Okay, so since it sounds like you were counting the clicks for periods of 5 minutes, then we can use Poisson (counting) statistics to understand the significance (or not) of the measurements. Under counting statistics, the uncertainty of a measurement of N counts can be calculated like this:     Uncertainty(N) = SquareRoot(N) And the uncertainty of the difference between two measurements N1 and N2 can be calculated like this:     Uncertainty(N2-N1) = SquareRoot(N1+N2)     [Note the plus sign on the right hand side!] So here's how I would report your results, including the excess:

Time since rainfall	Background counts	Sample counts	Excess counts
2:00 hours	95±9.7	103±10.1	8±14.1
6:20-7:00 hours (#1)	98±9.9	112±10.6	14±14.5
6:20-7:00 hours (#2)	92±9.6	106±10.3	14±14.1
6:20-7:00 hours (#3)	91±9.5	83±9.1	-8±13.2

Or you can divide everything by 5 to get the counts per minute:

Time since rainfall	Background (CPM)	Sample (CPM)	Excess (CPM)
2:00 hours	19.0±1.9	20.6±2.0	1.6±2.8
6:20-7:00 hours (#1)	19.6±2.0	22.4±2.1	2.8±2.9
6:20-7:00 hours (#2)	18.4±1.9	21.2±2.1	2.8±2.8
6:20-7:00 hours (#3)	18.2±1.9	16.6±1.8	-1.6±2.6

Since none of the excess counts are larger than twice the uncertainty, your measurements are inconclusive as to whether the sample is radioactive. The uncertainties also explain that anomalous fourth result -- it is lower than average, but not significantly so. So it is not an anomaly after all! Good point about the lead caves we use. In order to win the "signal-to-noise" problem, which is what you have stumbled upon, we try to get as few counts from background as possible by shielding with lead. We also use a much larger and more powerful detector than the unit you have, and we get a spectrum from the radiation instead of a "click." We also sometimes count the samples for days -- the counting time is very important! So all of that combines to give us very high sensitivity to radioactive material. By the way, I wanted to leave you with something to think about. Based on the background count rates you are getting, we can estimate your detector's sensitivity to a change in count rate -- sort of like your version of the Minimum Detectable Activity. Using an average background rate of 18.5 counts per minute, and counting statistics:

Counting time (minutes)	Count rate sensitivity (CPM) (95% confidence)
1	12.2
5	5.4
10	3.8
30	2.2
60	1.6

Also one further thing -- try to measure as soon after rainfall as possible. Every 40 minutes, the radon progeny activity decreases by about half! Mark [BRAWM Team Member]