Not assuming, in any way shape or form that the Mail hasn’t simply got it horribly wrong …

You need to remember the decay chains for various isotopes – therefore if there is contamination of a long-half substance (say U238), then you will get detectable amounts of many if not all of the decay products. Yes, 100g (say) of radon will have decayed through 9600 half-lives (i.e. there will be bugger all left). However, some of the decay products will have longer half-lives and many of these can be quite dangerous, both as toxins and as alpha-emitters, if ingested. You will probably still have more than 50g of radioactive material (more if the previously stable containers have been affected by alpha bombardment) and most of the remainder (say, taking it to 99.99g) will be generally icky.

Actually checking the decay series before I posted at the Speccie, the fly in the ointment is Lead 210, with a half-life of 22 years.

I was wrong above (mostly because Polonium 210 decays to the stable lead isotope 206) but from your 100g of Radon, you will have some 6-7g of radioactive Lead 210 left in amongst the Lead-206, and the latter will decay to Polonium 210 (via Bismuth 210) – so there will be readily detectable amounts of both of the latter hiding amongst the Lead 206 & 210.

dearieme

It’s rather odd that in a Psychology department there is apparently no-one who can do the statistical sums to tell us the probability of the observed cancer deaths being due to chance. Odd but not surprising?

Well, concerning radon, I do not think the reduction of radioactivity over the last 100 years is 10 to the power -10,000 but rather 2 to the power -9,611.8 (assuming you are correct on the half life of 3.8 days). [Actually, I read/calculate that 2 to the power -9,551.7 is more accurate.]

However, as it is a (bloody) gas, I do not think (unless there is some generative source in/at the laboratory) that this incorrectness of yours (overstating decay by around 3,006 times) is particularly relevant: it will all have blown out the (equally bloody) windows and door(s).

The polonium is a bit more of a problem, though I doubt much of one.

As SE has commented, half life, might not be sufficient knowledge, as the radiation might have activated secondary radiation in other substances, some of which have higher half lives, or even chemical toxicity (well, at least if one goes around licking the benches and floor, and assuming no significant cleaning per century).

So much for the danger of that specific location.

Then, and I suggest more importantly, one has to consider the other possible causes (and rates) of death from brain tumour and from pancreatic cancer. In the USA, there are approximately 13,000 annual deaths from brain tumour, from all causes: cancer and otherwise (UK equivalent 2,600). In the USA, there are approximately 34,290 annual deaths from pancreatic cancer (UK equivalent approx 6,850).

There are currently (so I count from their website staff list) 244 staff in MU’s psychology department. From a randomly selected group of 244 persons (over all ages) over the 36 years (the period since 1972), 0.38 persons, on average would die from brain tumours and 1.004 persons on average from pancreatic cancer. It is, at least for me, no great stretch of the imagination (or of the statistics) for 2 persons to die of each illness over the same period in such a group, without it being so very unlikely as to properly raise suspicions.

[Note. I’d love to carry on and work out the probabilities of 2 of each sort of death (and do know the maths for this). However, the worth of it all is a bit suspect for reasons of doubt about using, as above, just simple averages and about certainty of knowledge of the whole scenario. Firstly, the selected population is, presumably, all over 18 years of age: my assumption of background death rates have no 18+ component, which would increase the average death rate. Secondly, the reported deaths include at least one person (aged 69) who is presumably retired and outside the current staff population of 244, so a higher population count should probably be considered. Thirdly, and on the other side in effect, we have no certainty that the 2+2 records all deaths from the specified illnesses, over the whole 36 years. Nor do we know that the staffing levels have been at 244 for the whole 36-year period. And so on …]

Maybe the Health and Safety Executive will work it all out for us, with more accurate knowledge of the actual scenario, and also the ‘other’ radiation history of those who died. Or perhaps they will just trug around with a Geiger counter and then report no evidence of raised radiation levels now (for what value that has).

Still, it does look much more like ‘woo’ than anything else.

Best regards

gene berman

dearieme:

“cancer deaths due to chance.” Really?? Aren’t they all? Or none? Am I missing something?

Monty

At some point during the tenure of the building by the Physics department, there was probably an antecedent ( ie something which generates Radon through decay) of Radon on the premises. Radium perhaps.

And Radon generates some fairly unpleasant radioisotopes when it decays.

But what we do not know, is the composition of the bedrock under this part of the city. So we don’t even know if the level of Radon is higher than natural background levels.

Why doesn’t the Spectator bother to publish submitted comments? Any ideas, Tim? Just a shortage of minions with himself flaunting his locks in another place?

Tim adds: Quite possibly, yes….

dearieme

Gene, I was suggesting that the sums be done along the line pursued by Nigel: “….for 2 persons to die of each illness over the same period in such a group, without it being so very unlikely as to properly raise suspicions.” The question is whether the elevation in cancer death rate (if it is elevated) is small enough to be attributed to chance. Elevation by a factor of 2 is too small to be a worry; a factor of five might be worth investigating.

Fine comments all – but they aren’t worried about the right calculation.

Nigel Sedgewick is part way there. We could calculate the probability of getting the number of deaths observed from the rates of occurrence.

But, this only tells us the probability of the occurrence of a cluster in that location (not zero, but not very high either). The problem is that we need the probability of a cluster occurring anywhere – this is much closer to one.

Once such a cluster occurs and is found out by journalists, the next step is to look back selectively into the history of that single location for a culprit from the usual list of suspects: usually radioactive or chemical poisoning from either scientists or industrialists.

In sum, this is news because it follows a recipe that is virtually certain to produce a story often enough to be in the reporters toolkit, but infrequently enough that they can get away with this sort of nonsense.

Noting dearieme’s comment, I have crunched some numbers through the appropriate binomial distribution coefficient calculation for permutations and combinations. This is for the case of 244 persons, and probabilities over 36 years taken as those from the USA death rates in my earlier comment.

For the brain tumour case:
-probability of 0 deaths is 68.32%
-probability of 1 death is 26.05%
-probability of 2 deaths is 4.95%
-probability of 3 or more deaths is 0.69%

For the pancreatic cancer case:
-probability of 0 deaths is 36.57%
-probability of 1 death is 36.86%
-probability of 2 deaths is 18.51%
-probability of 3 or more deaths is 8.07%

Thus (the actually reported) 2 deaths from brain tumour would occur by chance in 4.95 percent of cases of randomly selected groups of that size. This is 13.8 times less frequently than for the commonest number of deaths (which is zero).

Likewise (the actually reported) 2 deaths from pancreatic cancer would occur by chance in 18.51 percent of cases of randomly selected groups of that size. This is 1.99 times less frequently than for the commonest number of deaths (which is one death, though zero deaths is nearly as likely).

Neither of these sets of figures, alone, is sufficiently rare to warrant further investigation. Combining them raises the further concern of cherry picking or of a ‘data dredge’, so I think any further investigation would be better to concentrate on statistics for all illnesses/deaths affected by atomic radiation.

I’ll just again draw attention to the various reservations from my previous comment, that the underlying assumptions made in these calculations are, in my view at least, not really good enough to rely on them overmuch. However, the actual method of calculation is the appropriate one.

So how many curies is 7g of Pb210 with a half life of 22 years? 7g = 1/30 mol ~ 2 × 10^22 nuclei
t1/2 = 22a ~ 6.94 × 10^8 s
exp(-λ t1/2) = 1/2
λ, specific activity = ln(2)/t1/2 ~ 10^-9/s
Curie = 3.7 10^10 Bq
So about 540 curies

Sounds like a lot, until you realise that Chernobyl released several tens of millions of curies, and continent-wide the excess cancer deaths are barely distinguishable from noise.

Radon is a gas so after a 100 years there would not be any left anyway – unless it is air tight.

Do they mean radium?

Not assuming, in any way shape or form that the Mail hasn’t simply got it horribly wrong …

You need to remember the decay chains for various isotopes – therefore if there is contamination of a long-half substance (say U238), then you will get detectable amounts of many if not all of the decay products. Yes, 100g (say) of radon will have decayed through 9600 half-lives (i.e. there will be bugger all left). However, some of the decay products will have longer half-lives and many of these can be quite dangerous, both as toxins and as alpha-emitters, if ingested. You will probably still have more than 50g of radioactive material (more if the previously stable containers have been affected by alpha bombardment) and most of the remainder (say, taking it to 99.99g) will be generally icky.

Actually checking the decay series before I posted at the Speccie, the fly in the ointment is Lead 210, with a half-life of 22 years.

I was wrong above (mostly because Polonium 210 decays to the stable lead isotope 206) but from your 100g of Radon, you will have some 6-7g of radioactive Lead 210 left in amongst the Lead-206, and the latter will decay to Polonium 210 (via Bismuth 210) – so there will be readily detectable amounts of both of the latter hiding amongst the Lead 206 & 210.

It’s rather odd that in a Psychology department there is apparently no-one who can do the statistical sums to tell us the probability of the observed cancer deaths being due to chance. Odd but not surprising?

Well, concerning radon, I do not think the reduction of radioactivity over the last 100 years is 10 to the power -10,000 but rather 2 to the power -9,611.8 (assuming you are correct on the half life of 3.8 days). [Actually, I read/calculate that 2 to the power -9,551.7 is more accurate.]

However, as it is a (bloody) gas, I do not think (unless there is some generative source in/at the laboratory) that this incorrectness of yours (overstating decay by around 3,006 times) is particularly relevant: it will all have blown out the (equally bloody) windows and door(s).

The polonium is a bit more of a problem, though I doubt much of one.

As SE has commented, half life, might not be sufficient knowledge, as the radiation might have activated secondary radiation in other substances, some of which have higher half lives, or even chemical toxicity (well, at least if one goes around licking the benches and floor, and assuming no significant cleaning per century).

So much for the danger of that specific location.

Then, and I suggest more importantly, one has to consider the other possible causes (and rates) of death from brain tumour and from pancreatic cancer. In the USA, there are approximately 13,000 annual deaths from brain tumour, from all causes: cancer and otherwise (UK equivalent 2,600). In the USA, there are approximately 34,290 annual deaths from pancreatic cancer (UK equivalent approx 6,850).

There are currently (so I count from their website staff list) 244 staff in MU’s psychology department. From a randomly selected group of 244 persons (over all ages) over the 36 years (the period since 1972), 0.38 persons, on average would die from brain tumours and 1.004 persons on average from pancreatic cancer. It is, at least for me, no great stretch of the imagination (or of the statistics) for 2 persons to die of each illness over the same period in such a group, without it being so very unlikely as to properly raise suspicions.

[Note. I’d love to carry on and work out the probabilities of 2 of each sort of death (and do know the maths for this). However, the worth of it all is a bit suspect for reasons of doubt about using, as above, just simple averages and about certainty of knowledge of the whole scenario. Firstly, the selected population is, presumably, all over 18 years of age: my assumption of background death rates have no 18+ component, which would increase the average death rate. Secondly, the reported deaths include at least one person (aged 69) who is presumably retired and outside the current staff population of 244, so a higher population count should probably be considered. Thirdly, and on the other side in effect, we have no certainty that the 2+2 records all deaths from the specified illnesses, over the whole 36 years. Nor do we know that the staffing levels have been at 244 for the whole 36-year period. And so on …]

Maybe the Health and Safety Executive will work it all out for us, with more accurate knowledge of the actual scenario, and also the ‘other’ radiation history of those who died. Or perhaps they will just trug around with a Geiger counter and then report no evidence of raised radiation levels now (for what value that has).

Still, it does look much more like ‘woo’ than anything else.

Best regards

dearieme:

“cancer deaths due to chance.” Really?? Aren’t they all? Or none? Am I missing something?

At some point during the tenure of the building by the Physics department, there was probably an antecedent ( ie something which generates Radon through decay) of Radon on the premises. Radium perhaps.

And Radon generates some fairly unpleasant radioisotopes when it decays.

But what we do not know, is the composition of the bedrock under this part of the city. So we don’t even know if the level of Radon is higher than natural background levels.

Why doesn’t the Spectator bother to publish submitted comments? Any ideas, Tim? Just a shortage of minions with himself flaunting his locks in another place?

Tim adds: Quite possibly, yes….

Gene, I was suggesting that the sums be done along the line pursued by Nigel: “….for 2 persons to die of each illness over the same period in such a group, without it being so very unlikely as to properly raise suspicions.” The question is whether the elevation in cancer death rate (if it is elevated) is small enough to be attributed to chance. Elevation by a factor of 2 is too small to be a worry; a factor of five might be worth investigating.

Fine comments all – but they aren’t worried about the right calculation.

Nigel Sedgewick is part way there. We could calculate the probability of getting the number of deaths observed from the rates of occurrence.

But, this only tells us the probability of the occurrence of a cluster in that location (not zero, but not very high either). The problem is that we need the probability of a cluster occurring anywhere – this is much closer to one.

Once such a cluster occurs and is found out by journalists, the next step is to look back selectively into the history of that single location for a culprit from the usual list of suspects: usually radioactive or chemical poisoning from either scientists or industrialists.

In sum, this is news because it follows a recipe that is virtually certain to produce a story often enough to be in the reporters toolkit, but infrequently enough that they can get away with this sort of nonsense.

Noting dearieme’s comment, I have crunched some numbers through the appropriate binomial distribution coefficient calculation for permutations and combinations. This is for the case of 244 persons, and probabilities over 36 years taken as those from the USA death rates in my earlier comment.

For the brain tumour case:

-probability of 0 deaths is 68.32%

-probability of 1 death is 26.05%

-probability of 2 deaths is 4.95%

-probability of 3 or more deaths is 0.69%

For the pancreatic cancer case:

-probability of 0 deaths is 36.57%

-probability of 1 death is 36.86%

-probability of 2 deaths is 18.51%

-probability of 3 or more deaths is 8.07%

Thus (the actually reported) 2 deaths from brain tumour would occur by chance in 4.95 percent of cases of randomly selected groups of that size. This is 13.8 times less frequently than for the commonest number of deaths (which is zero).

Likewise (the actually reported) 2 deaths from pancreatic cancer would occur by chance in 18.51 percent of cases of randomly selected groups of that size. This is 1.99 times less frequently than for the commonest number of deaths (which is one death, though zero deaths is nearly as likely).

Neither of these sets of figures, alone, is sufficiently rare to warrant further investigation. Combining them raises the further concern of cherry picking or of a ‘data dredge’, so I think any further investigation would be better to concentrate on statistics for all illnesses/deaths affected by atomic radiation.

I’ll just again draw attention to the various reservations from my previous comment, that the underlying assumptions made in these calculations are, in my view at least, not really good enough to rely on them overmuch. However, the actual method of calculation is the appropriate one.

Best regards

So how many curies is 7g of Pb210 with a half life of 22 years? 7g = 1/30 mol ~ 2 × 10^22 nuclei

t1/2 = 22a ~ 6.94 × 10^8 s

exp(-λ t1/2) = 1/2

λ, specific activity = ln(2)/t1/2 ~ 10^-9/s

Curie = 3.7 10^10 Bq

So about 540 curies

Sounds like a lot, until you realise that Chernobyl released several tens of millions of curies, and continent-wide the excess cancer deaths are barely distinguishable from noise.