Sunspots and Norwegian Child Mortality

Guest Post by Willis Eschenbach

In January there was a study published by The Royal Society entitled “Solar activity at birth predicted infant survival and women’s fertility in historical Norway”, available here. It claimed that in Norway in the 1700s and 1800s the solar activity at birth affected a child’s survival chances. As you might imagine, this attracted much interest, particularly among the credulati. So I wanted to take a look at their data.

Unfortunately, the authors did not archive the data and code. So I got in touch with said authors, and they kindly provided part of the data. This is a dataset of some 5,167 individuals showing birth years, sex, solar “MAX/MIN” category, and lifespan. In particular, my thanks to Professor Eivin Røskaft who has been most responsive to requests and queries from myself and others regarding the data and code.

Now, before I do any detailed analysis of a given dataset, I first want to see what all of the data looks like. It’s a graphical game, because we can’t gain anything by looking at a huge pile of numbers. We have to convert the numbers into patterns in order to grasp their nature, distribution, and significance. Before doing any detailed analysis I like to start by looking at the big picture, all of the data. So here was my first plot of the Norwegian data. It’s a plot of the lifespans of the individuals (vertical axis) versus their year of birth.

norwegian scatterplot age at death vs birth yearFigure 1. Lifespan by Birth Year, Norwegian data. In order to show the underlying structures, I use partly transparent dots. In that way, it’s darker where the data piles up. 

Mmmm … you can see why the first step should always be to look at the whole dataset. It’s obvious that we are looking at some curious combination of information which has led to two very different subsets of data. There are a bunch of people who died before they were twenty-one … and then there are a bunch of people who mostly lived a full lifespan, sixty to ninety years or so.

I have no idea why the dataset is divided in that manner. I thought it might be data from two different sources, but that turns out not to be the case. Professor Røskaft was kind enough to provide us with the following answer to my question:

We see your point about variation in sample sizes and age classes across years, and there are certainly many caveats when analysing such historical records. However, the data are all gathered from the same source, and the “two-datasets” impression is an inevitable consequence of the data collection. These data were collected with tracking individuals across generations in mind. Starting with the first church record, we could find the first adult generation, which we then used to track their offspring, which of course only were born ca. 20 years later. If you do a simple exercise of reading a church record, you will find births, marriages and deaths. Let’s say that we start with the first book in 1700. Then we will know the number of births that year, the number of marriages and number of deaths. For the married couples, we could even go back in time, since we know their age at marriage (let’s say they were born in 1675). Hence, the different distribution of the “two-datsets” is the difference between the generations. However, there are an equal NUMBER of generations in each group.

The lack of deaths around 20 years of age is probably caused by the fact that we could only track married couples staying in the same village as they were born. As stated in the note under Table S1, unmarried and emigrated individuals could not be included.

While I appreciate his reply, I still don’t see why any of those would make the sharp cutoff at exactly twenty years seen in Figure 1. I certainly may be missing something in his explanation. But let’s set that aside, and continue with the data.

As is common in my life, I’m looking at some kind of dataset that I’ve never analyzed before. So the diagonal lines in Figure 1 puzzled me for a bit. But then I realized that these were mass mortality events, likely epidemics or crop failures. And in fact there are records of a bad epidemic in Norway in 1772-3. This is the darkest of the diagonal lines in Figure 1. It starts at the bottom of the graph in 1773 and goes up to the left, as indicated by the red arrow. It’s described in one church record as a “blistering fever”, so perhaps measles, cowpox, or smallpox.

I bring this epidemic up to highlight a problem with the data. Look at how many infants, children and young adults died as a result of the 1773 epidemic, lots of deaths right up to the age of 20. And there are also deaths among older individuals, in their thirties and forties and more … but we have no recorded deaths of 21- or 22- or 23-year-old individuals. The recorded deaths cut off abruptly at exactly 20 years. So obviously the dataset is not an accurate representation of the deaths of the times … worrisome.

In any case, here’s my first cut at comparing the sunspot numbers at birth with the lifespans for the full dataset. Figure 2 shows a scatterplot of lifespan versus sunspots during the birth year.

norwegian scatterplot age at death vs sunspotsFigure 2. Scatterplot, lifespan versus sunspots during the birth year.

At least for the full dataset, there is no significant relationship between sunspots and lifespan.

Now, regarding the appearance of there being two datasets, Professor Røskaft was good enough to explain to us that it doesn’t matter because their work is based on infant and child mortality rates, viz:

As I am sure you know from reading our paper, we did not do any statistical analyses of life expectancy. For the survival analyses, we only used the subset of reaching adulthood or not (20 years of age). The life expectancies reported for each group in Table S1 (online) are merely means +/- SD of our complete dataset, as stated in the legend. As evolutionary biologists, our focus is fitness (reproductive success) and then you basically need to reach adulthood and produce many children. If you look at the survival curves (Fig S1), the major difference lies within the first two years of life. As we argue in our paper, the reduced lifespan is solely due to a high infant mortality.

So it appears that they are only using the lower section of the data shown in Figure 1, the under-21 data. OK, fair enough, that’s legit. There is a remaining problem, however. This is that nature is naturally clumpy. As the poet had it:

Shake and shake

The ketchup bottle

None’ll come and

Then a lot’ll.

The deaths of infants and children in Norway are a great example of this clumpy nature of nature. Figure 1 shows that at the longer term, there were lots of epidemics and/or crop failures from 1750 to 1800. Look at all of the slanted lines in the lower part of Figure 1 during that time. And then there were few mass mortality events for the next fifty years. But then from 1850 to 1900, the death rates jump again. Clumpy.

And the same is true on all time scales, from daily to annual to centennial. No deaths for a few years, or decades, and then a bunch. And this leads to a big problem when you want to relate and compare a clumped dataset to a highly cyclical phenomenon like say sunspots. The odds go way up that what looks significant is actually just random. When you are comparing a strongly cyclical signal like sunspots to a “clumpy” natural variable, it only takes a couple of clumps in the right spots to give the appearance of causality and correlation.

As a result of all of these considerations about just the dataset itself, I’d be very cautious about any interpretation of results from this dataset without a good hard look under the hood at whether they are significant. And with those caveats, let’s look at the under-21 lifespans. (I have omitted the final 20 years of the under-21 dataset, because we need at least 20 years of data to see the full lifespan distribution up to 20 years of age.)

Now the authors have divided the dataset into “MAX” and “MIN” based on solar activity during the birth year. I disapprove of this procedure for two reasons. First, I dislike it because it is arbitrary and subjective. For example, in this case the division ends up 70/30, with 70% of the individuals being counted in the “MIN” category, and 30% in the “MAX” category. But some other arbitrary threshold could reverse those numbers. So the binary “MAX/MIN” conversion is highly dependent on the threshold chosen, and with different thresholds you can get very different outcomes.

Second, it doesn’t make sense to me to replace a continuous variable (average annual sunspots) with a categorical variable, particularly a binary categorical variable (e.g. MAX/MIN). Seems to me like you’re throwing away valuable information when you do that. In particular you lose all dosage/response information. And when you are looking to establish causation, dosage/response is an important part of the evidence.

But that’s what they did, so let me do the same. Figure 3 shows two histograms of the lifespans of the under-21 part of the Norwegian dataset. The blue histogram is of the individuals listed as “MAX” for the sunspots during their birth year, and the hatched red histogram is of the “MIN” individuals. I’m using the authors’ MAX/MIN division as specified in the data they sent me.

hist norwegian under-21 lifespans minimum maximumFigure 3. Histograms, Norwegian under-21 lifespans during “minimum” and “maximum” solar activity. The solid blue columns show the individuals born during solar “maximum” (n = 502 individuals). The red hatched columns show the solar “minimum” individuals (n = 1,321 individuals).

Figure 3 shows what percentage of the individuals died at 0 years, 1 year, 2 years, and so on. For example, in both MAX and MIN groups about 38% of the individuals died before their first birthday.

In fact, there is very little difference at any age between the MAX and the MIN groups. I’m sorry, but I’m just not seeing any evidence. There’s no statistical difference between the survival rates based on the level of solar (sunspot) activity at birth.

The authors say that they have “controlled” for various other factors. According to their paper these factors are sex, maternal effects, socioeconomic status, cohort and ecology. I haven’t been given access to that data, so I can’t comment on their claim.

But when you start splitting things up into sub-sub-sub categories, you need to adjust your statistics to allow for that, and I can’t find any evidence that they’ve done so, or that they are even aware of the issue.

The problem is that if you look long enough you’ll find unusual things. Here’s an example. If a person takes six coins and throws them in the air at once, there’s only one chance in sixty-four that all six of them will fall heads up. That would be unusual. That result would be called “statistically significant”.

But if she picks all six coins back up and tosses them again, and does the same again, and again, sooner or later they’ll come up six heads. And when they do so, the person can’t claim at that point that six heads is statistically significant. It’s not. Six heads is unusual, but it’s only unusual if the coins are just tossed one time. It’s not unusual if the six coins are flipped twenty times. And each subdivision of your data, say first into male/female and then into max/min, is the same as flipping the coins twice as many times …

Note how far the subdivision has gone in their analysis. Above, I show all individuals divided into two groups by solar activity, MAX and MIN. The next step would be to split them up by sex, so then we have four groups (max and min male, max and min female). Divide them further into high and low socioeconomic status and we have eight groups. Then dividing by ecology gives us sixteen groups, and an additional division by cohort gives us thirty-two groups …

With 32 groups, are we surprised that we find things that look significant but aren’t? With thirty-two tosses of six coins, you have about a 40% chance of finding a toss that’s six heads … with that fine a division it’s not significant at all to find six heads.

As a result, I fear they have not established their claims. They haven’t allowed for the number of trials in their calculation of statistical significance.

Let me show one final look at just the under-21 data. Figure 4 shows a scatterplot of lifespan versus sunspot count for just the under-21 part of the Norwegian dataset. It’s the same as Figure 2, but for the under-21 individuals. This avoids the whole “MIN/MAX” problem, and relates the two variables directly:

scatterplot lifespan vs sunspot norway under 21Figure 4. Scatterplot of the lifespan of the individuals in the under-21 Norwegian dataset, versus the number of sunspots during the birth year. Red line is the trend line of lifespan with sunspots.

Well, there you have it. When the trend line is horizontal as in Figure 4, that means that there’s no relationship between under-21 lifespan and sunspots. Nor does this surprise me. If sunspots had that much effect on human gestation and survival, we’d see the same effects all over the map in all kinds of living organisms … but we don’t.

So are the findings of the authors statistically significant? I don’t know, because the authors haven’t provided me with the data to test their findings, and because the authors aren’t aware of the problems with their significance calculations.

But given my findings in Figures 4 & 5 regarding the full under-21 dataset, I greatly doubt the significance of their findings regarding sub-sub-subsets of that data. Even if we were to find a putative effect on say boys, who were high-status, from a particular cohort, born during the MIN, and raised in a certain ecology, it would have to be a very, very strong effect to be significant.

In closing, I have to give a mixed report regarding the authors. In their great favor is the fact that they were responsive to requests from myself and from others for data and code. Professor Røskaft in particular was most forthcoming, answering questions and providing further data. So all of the authors have my thanks for that.

However, at the end of the day I still don’t have their full dataset with all of the status and ecology and other information … it’s another example of why providing the data as used and the code as used is so important.

Best to all,

w.

My Usual Request: If you disagree with someone, please have the courtesy to quote their exact words that you disagree with. In that way, we can all understand the exact nature of what you object to.

Further Reading: After writing much of the above, I found that in some aspects I’d been anticipated by the irrepressible Matt Briggs, Statistician To The Stars.

Data: The Norwegian Data provided by Professor Røskaft is here as a text file.

Advertisements

99 thoughts on “Sunspots and Norwegian Child Mortality

  1. I read of a similar assertion a couple years back in an epi-genetics paper (generational diabetes I think). The correlation wasn’t to sunspots, rather it was to food production. There was more disease and starvation in poor harvest years, ergo higher death rates…. seems reasonable I guess.
    I’ll see if I can find it. Sunspot number may correlate with harvest yields in high latitude locations?

    • Every economist has heard of Jevons’s Sunspot Theory (dating from the 19th century). He assumed that trade or business cycles were caused by sunspot activity (assuming that the latter influenced weather and thus harvests).
      When I first learned of this it seemed like a very bizarre theory. These days “sunspots” in economics have a very different meaning, namely “extrinsic random variables upon which participants coordinate their decisions”.

    • Willis
      Whether their study makes any sense or not, the the 2 50 year periods you mention are of intrest to me 1750 to 1800 and 1850 to 1900. can we tie any climatic events to these to periods? Are these evedence of previous warming/cooling cycles or are they just anomalies in the data?

  2. “The lack of deaths around 20 years of age is probably caused by the fact that we could only track married couples staying in the same village as they were born.”
    Not necessarily. Children are born and reared in the village, but some of the young adults move away. Their deaths abroad would prevent the authors from counting them. Some of the now-middle-aged adults may move back to their villages, allowing the authors to include their middle-aged or elderly deaths, without counting the young-adult deaths abroad, or at sea, or in the military. Norwegians were famously sailors, so this fits. Difficult to test, though.

    • Lots of Norwegians emigrated to America also. Many probably in their 20’s.
      I found this http://www.borgos.nndata.no/migrants.htm
      “During the hundred years between 1825 and 1925 many Norwegians left their homes and emigrated to America. The exact numbers may have been between 800 and 900 thousands. Compared with the population size Norway comes second to Ireland in sending emigrants to America.”
      The total population in 1825 was 1,051,318 and in 1900 2,240,032
      http://en.wikipedia.org/wiki/Demographics_of_Norway
      If 8-900 000 left for America during that period, for sure it would show up as “missing deaths at home” among people in their 20’s.
      Btw. I am Norwegian

      • I am Norwegian too, Amatør 1, and I just want to ask where I can buy carbon credits in order to reduce the terrible child mortality in 18th century Norway? Maybe Al Gore can help?

      • “…I just want to ask where I can buy carbon credits in order to…”
        ConTrari, I still have some carbon cred left, but they are going quickly causing the price to rise.
        We must save the children! Cash will do, unless you have a PayPal account?

      • On a related matter I remember one paper on the stature of Northern European Men during the Medieval Warm Period compared to LIA.

        Abstract
        New Light on the “Dark Ages”
        The Remarkably Tall Stature of Northern European Men during the Medieval Era
        Based on a modest sample of skeletons from northern Europe, average heights fell from 173.4 centimeters in the early Middle Ages to a low of roughly 167 centimeters during the seventeenth and eighteenth centuries……
        http://ssh.dukejournals.org/content/28/2/211.short

        After LIA we Europeans grew taller again. Cols stunts growth! / sarc. 😉

    • Years ago I saw a chart in the Viking Museum in Oslo, which showed how the population of Norway varied over the years. The Chart suggested that bad weather (cold and damp) led to famines, which led to mass deaths. To the Norwegian readers, anyway to verify whether this chart is still in the museum. Interestingly, the museum also had extensive exhibits on how the ice age ended the Viking era.

  3. I’ve had at least 4 or 5 times in my life, where a different split-second decision would have cost me my life.
    I guess it was all in the sunspots, not my reactions or blind luck.
    Seeing as I’m still pecking away at this keyboard, the sunspots must be in my favor.
    Who knew ?

  4. Willis, is your final scatterplot normalised for population vs sunspot count? Some sunspot counts might occur less frequently than other sunspot counts, which might bias the weight of the scatter in favour of more frequent sunspot counts.

  5. Seems to me solar activity at time of conception would be more influential than time of birth.

    • If the sunspots were affecting crop yields, eg sudden cold periods producing crop failures, then the sunspot activity would be linked to the year of death, not birth.

  6. Seems to me the since the claim is sunspot number affects child lifespans, a line should be drawn at the age marking the end of childhood (or infancy) and compared with lifespans above and below using the Number Spots vs. Lifespan plots. Groupings along the x-axis should be avoided but perhaps in ranges of +/- 25 spots or so.
    The percentages below the line for a given could be compared against percentages for other columns; As usual, one needs to determine if the resulting comparisons are predictive. A leave-N out approach doesn’t appear to be feasible with these data.

      • Hmmmm…. LOL!
        It’s not at all clear what Willis has done but the under-21 plot above doesn’t seem to convey much.
        (1) the actual age at death is not important so we should be looking for trends in lifespan percentages above and below a threshold and not lifespan itself since we are interested in childhood vs. non-childhood status;
        (2) there is no justification really for throwing out lifespans longer above 20; and
        (3) as ever, p-value hypothesis testing is meaningless as it is a reflection of the Pr(parameters | model, data) and not Pr(model | parameters, data) — how well the knobs have been twisted says nothing about usefulness.

  7. As you might imagine, this attracted much interest, particularly among the credulati.
    I have eclectic interests and always thought I have a vivid imagination. Unless there is some significance in the article that escapes me, I can’t imagine why this is of interest to anyone. My interpretation (perhaps Mis interpretation?) is that this is filler while we wait for something of significance to happen in the “climate change” world. I do, none-the-less, appreciate Mr. Eschenbach’s contributions to WUWT.

    • Isn’t it interesting that it attracted much interest?
      Did you click on the link?
      It’s worse than we thought.

    • @PMHinSC: sun spots – climate change – death rates —-> all the requirements for a big media storm over global warming caused by CO2 (not sure how CO2 fits, but that will not stop the media!).
      However, this article is about data analysis. We have some records, and we are indebted to the authors of this paper for collating said data, and on request, putting it into the public domain.
      The authors have done an analysis and written up their findings. Willis has analysed their results with restricted access, and summarised his results here.
      The significance? Climate research is all about data analysis. collect large amounts of data and bat back and forth a few times to see what it can show you.
      I suggest this article demonstrates that the same data set can be analysed in a number of ways, results presented in a number of ways, leading to opposite conclusions.
      This is mirrors what happens in climate research all the time.

  8. How many solar eclipses or lunar eclipses occurred during that period? How many of those were total in Norway? How many Hawaiian volcanoes erupted during that time. I think that total eclipses and Hawaiian volcanoes might be better correlated to likelihood of living to age 20 in Norway than sunspots are. There isn’t any reason why Hawaiian volcanoes should have anything to do with life in Norway, but then again there isn’t reason why sunspots should have anything to do with it either. By the way a butterfly flapping its wings can’t cause a hurricane.

  9. I would have expected a 1783-84 Laki volcano mortality effect too. Tough vikings. The microbes get them, while they laugh at Surtr.

  10. I love your “clumpy nature of nature”.
    These are your exact words and per your reasonable request to always use your exact words, here is the challenge for you about “clumpy nature of nature” relevant to climate change (read global warming).
    Below is the most up to date graphical presentation (up to January 2015) of what I think Mother Nature generated: three clumps of global warming importance (I was going to say significance). I think this is the absolute best up to date presentation of temperature vs. CO2 data regarding modern times (read period influenced by burning fossil fuels) showing 3 individual clumps. My interpretation of this graphic presentation is that the first clump presents no evidence of temperature increase or decrease. The second clump presents very good evidence that temperature was increasing nicely with CO2 increases. The third clump presents no evidence of temperature increase or decrease. I trust you can find the clumps to justify your suspicion (or theory) regarding the “clumpy nature of nature”
    Climate 4 you temperature anomalies vs. CO2 graphical presentation up to January 2015.
    http://www.climate4you.com/
    “To investigate the potential significance of this visual impression, all monthly HadCRUT3 temperatures were plotted against the monthly Mauna Loa measurements in the diagram below.”
    (To get the diagram, you need to copy the link below and then paste it in your search engine or go to the climate4you site and get it from there, I tried posting it here but was unable to, then if you want to do your own “clump” analysis you would have to download the corresponding temperature and CO2 concentrations data files from the respective sites and this I know you can easily do)
    http://www.climate4you.com/images/HadCRUT4%20GlobalMonthlyTempSince1958%20VersusCO2.gif
    Below is the interpretation of the diagram at climate4you site. I am not in agreement with their interpretation of the first and third clump but I agree with their interpretation of the second clump. It would be interesting to have your interpretation of these clumps, using the statistical procedure(s) of your choice and even possibly you could find more than 3 clumps.
    “Diagram showing HadCRUT4 monthly global surface temperature estimate plotted against the monthly atmospheric CO2 content according to the Mauna Loa Observatory, Hawaii, back to March 1958. The red line is a polynomial fit with key statistics listed in the upper left part of the diagram. Last month shown: January 2015. Last diagram update: 2 March 2015.
    The diagram above shows all HadCRUT3 monthly temperatures plotted against the monthly Mauna Loa CO2 values, since the initiation of these measurements in 1958. As the amount of atmospheric CO2 have risen steadily since 1958, although with annual variations, the oldest values of temperature and CO2 are plotted close to the left side of the diagram, and more recent values are progressively plotted towards the right side of the diagram.
    By this, the diagram illustrates that the overall relation between atmospheric CO2 and global temperature apparently has changed several times since 1958.
    In the early part of the period, with CO2 concentrations close to 315 ppm, an increase of CO2 was associated with decreasing global air temperatures. When the CO2 concentration around 1975 reached 325 ppm this association changed, and increasing atmospheric CO2 was now associated with rising global temperatures. However, when the CO2 concentration at the turn of the century reached about 378 ppm, the association changed back to that characterizing the period before 1975. Thus, since 2000, increasing concentration of atmospheric CO2 has again been associated with decreasing global temperature.
    The diagram above thereby demonstrates that CO2 can not have been the dominant control on global temperatures since 1958. Had CO2 been the dominant control, periods of decreasing temperature (longer than 2-5 years) with increasing CO2 values should not occur. It might be argued (IPCC 2007) that the CO2 dominance first emerged around 1975, but if so, the recent breakdown of the association around 2000 should not occur, either.
    Consequently, the complex nature of the relation between global temperature and atmospheric CO2 since at least 1958 therefore represents an example of empirical falsification of the hypothesis ascribing dominance on the global temperature by the amount of atmospheric CO2. Clearly, the potential influence of CO2 must be subordinate to one or several other phenomena influencing global temperature. Presumably, it is more correct to characterize CO2 as a contributing factor for global temperature changes, rather than a dominant factor.
    The breakdown of the positive temperature-CO2 relation since about 2000 (diagram above) have now lasted 10-11 years. This suggests that the recent global temperature development might deviate significantly from previous short-lived (2-5 years) periods of cooling derived from oceanic and volcanic activity as seen several times between 1975 and 2000. There are two possibilities: 1) Global air temperatures may again begin to increase in a short while. 2) The recent development may represent the beginning of a more thorough and long-lasting cooling, perhaps similar to the cooling period after 1940. As usual, time will show what is correct.”

    • Cool. I was not able to see that the diagram would be automatically displayed. Now I see it. Nice.

    • I don’t see any “three clumps” in the graph, I rather see result of nonlinear process.
      I think it could be supplemented by this sophisticated wiggle match:
      http://www.woodfortrees.org/plot/hadcrut4gl/from:1959/to:2013/mean:12/detrend:0.6/offset:0.4/plot/esrl-co2/from:1959/to:2013/derivative/mean:12/mean:12/scale:2
      which is strongly suggesting that temperature plays significant role in CO2 increase rate but the effect diminishes at higher CO2 concentrations.
      I doubt there’s anything unknown on that, the question is what is the right interpretation.

      • the question is, what is the right interpretation?
        A: For many climate modelers, it’s the one that offers the highest probability of maintaining funding levels from the governments. Natural variability just does not make the cut.

    • One problem with this plot is tha it uses monthly values of historical temp anomalies vs. monthly CO2, but then places a time line (1958-2015) above the CO2 values. The inter annual variation of CO2increases in NH winter early spring then decreases from about mid-May into the fall. Last time I checked though, time does not run backwards (not that we’d be aware of it if it did,).

      • You can see how the annual averages CO2 concentrations increases with time from the plot at the Mauna Loa site. The lowest CO2 concentrations are reported in 1958 and highest in 2015.
        Simply, CO2 increases every year. I would prefer using yearly averages but we would get the same pattern.
        Obviously the advantage of monthly is the much larger number of data points available for regression analysis but then we get the seasonal variation as you mentioned.

    • My suggestion would be to plot temperature anomaly on the x-axis ( as the independent varible), and thus CO2 as the dependent varible.
      Observed CO2 levels are driven by the competition between carbon sinks and carbon sources in our oxidative biosphere. There is no reason not to think that in a cooling but warm world with high photosynthetic activity due to heightened pCO2, that annual CO2 levels will not fall. Thus CO2 is the dependent variable as it can and will eventually fall with time regardless of anthropogenic input, which is trivial in relation to natural sources.

      • This is not the claim of the IPCC! The claim is that the increase in atmospheric CO2 is the result of burning fossil fuels and therefore WE are responsible for increasing global temperature. We must stop burning.
        What we are seeing now is CO2 is still increasing but temperature is not following, as shown on the graph.

      • I understand what the IPCC claims. Those claims are based on the circular logic that the climate modeling community used to tune the model parameters and produce the GCM ensembles.
        I fully realize anthropogenic CO2 is a source of pCO2. But there are many natural sources of CO2, the sum of which dwarfs human output on an annual scale. A simple measured pCO2 value (at Mauna Loa) is a simplified snapshot of dynamical gas measurement, with constantly changing rates of release from many sources (natural and manmade), and changing rates of absorptions (natural) from many sinks. The balance of which for the last 200 years or so have been for steadily increasing pCO2. There is good reason to believe that at least 40% of this pCO2 rise has been due to man’s steadily increasing burning of all sources of carbon fuels. The other portion is due to man’s forced changing land use (agriculture, deforestation), and also a natural warming world (coming out of the LIA).
        But it is very likely that ∆CO2 lags ∆T by some period of decades as the dynamical inflows and outflows change. Not the other way around.
        Thus: ∆(pCO2) ≈ ∂T/∂t. and not ∆T ≈ ∂(pCO2)/∂t .
        But labeling pCO2 a proxy for time (on the same axis) is simply wrong regardless of which way nature eventually tells us is the correct direction of causality.

      • I understand your last reply of your understanding of the IPCC claim.
        However if we let this claim by the IPCC go we (industrialized countries) and other countries like India and Africa will have a very high price to pay. We are being accused of creating global warming. If true, well indeed we must change our way, regardless of the cost. However, the IPCC never directly produced evidence. They only showed temperature increase data separately from CO2 increase data. Indeed we can see from the plot above that for the middle period there is a correlation between the two and it is a justification to pursue CO2 as a driving force, but then this correlation fades as CO2 increases continued but not so for temperature.
        It is not the first time that plots of CO2 vs. temperature have been made with a variety of approaches showing different patterns. But now we are beginning to have a sufficient number of years (after 1959) to work with and no longer have to go prior to 1959. The authors of this graph, I am sure know what they are doing using monthly data instead of annual data. I think we will see this graph updated every month at their site! They are willing to wait (and so do I) but waiting a month is better than a year!
        I took your suggestion of looking at yearly instead of monthly averages this afternoon. I don’t have the HadCRUT4 data on my computer at this time. So I plotted the GISS yearly temperature data vs. the CO2 yearly concentrations. Using either yearly CO2 concentrations or year number since 1959 vs. GISS temperature anomalies and the same polynomial analysis, I get the same results as shown in climate4you graph shown above. In fact I get a little better R squared value (0.89, can you believe) than they did, obviously the annual average shows less variation (as you pointed out) than monthly due to seasonal changes. The advantage on monthly, well is just faster update available.
        I am really looking at this from a practical point of view. Simply, is the IPCC claim correct.
        I am open to any more suggestions.

  11. Further to the idea that sunspots have something to do with health and mortality, from Dr. John Cannell et al Epidemic Influenza and Vitamin D (2006) which is worth reading in its entirety (as is the 2008 followup article):
    Hope-Simpson was the first to note an association between severe influenza epidemics and solar flare activity [66]. In 1990, Hoyle and Wickramasinghe confirmed the association but von Alvensleben disputed it [67, 68]. Horgan [69] promptly derided the observations, connecting them to viral invasions from outer space, a theory Hope-Simpson dismissed in his 1992 book [3]. Since the controversy, science has learned that solar flare activity increases high-altitude ozone, which, in turn, absorbs more UVB radiation thereby decreasing surface UVB [70]. Thus, paradoxically, heightened solar activity reduces surface UVB; presumably, average 25(OH)D levels would be lower as well. Rozema et al. estimated the variations in surface UVB radiation due to the solar flare activity over the last 300 years and estimated that, beginning in the eighteenth century, ‘the dose of surface UV-B should be (about) 4% to 13% lower at maxima of the 11-year solar cycle’ [71]. Although modest, such reoccurring decreases in UVB radiation should trigger reductions in average 25(OH)D levels, which, in turn, could trigger nonlinear factors related to influenza infectivity.
    http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2870528/
    The 2008 article is On the Epidemiology of Influenza
    http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2279112/

    • More from Cannell about Norway and, as a bonus, ENSO!
      If vitamin D is Hope-Simpson’s ‘seasonal stimulus’, then countries with low 25(OH)D levels and marked wintertime troughs should have higher excess wintertime mortality than do countries with high 25(OH)D levels and little seasonal variation. For example, Norway has the highest 25(OH)D levels in Europe (thought to be due to its high year-round consumption of fish and cod liver oil) [59]. Levels of 25(OH)D in Scandinavia display the least seasonal variation in Europe; indeed there is virtually no 25(OH)D seasonal variation among the elderly in Scandinavia [60]. On the other hand, the elderly in Great Britain have low 25(OH)D levels and such deficiencies are much more common during the influenza season [61]. Excess wintertime mortality is twice as high in Great Britain as in Norway [62].
      Global weather changes are associated with El Niño/Southern Oscillation (ENSO) [63]. Viboud et al. found an average of 3·7 million influenza cases in France during the 10 cold phases of ENSO but only 1·8 million cases during the eight warm phases [64]. The same authors reported that cold ENSO phases are associated with colder temperatures in Europe. Colder temperatures should lower mean serum population 25(OH)D levels by lessening outdoor activity and necessitating more clothes when outdoors. Ebi et al. studied six Californian counties and found that hospitalizations for viral pneumonia peaked around the winter solstice in all six counties [65]. They also found hospitalizations increased 30–50% for every 5 °F (3 °C) decrease in minimum temperatures in four counties and increased 25–40% for every 5 °F (3 °C) decrease in maximum temperatures in the other two.

      • Excess wintertime mortality is twice as high in Great Britain as in Norway

        In my opinion, this has to do with how houses are built, and probably a few other things. Building standards are not comparable. Norwegian homes are warm, we don’t live in caves.

      • For the life of me, I don’t see how any rational person can d*ny the obvious influence of solar activity on ENSO.

      • In Norway we eat pizza, like every other civilized society, and we hate cod liver oil. And as Amatør 1 writes, our houses are warm and well-built. We can not have the luxury of chillblains and droughty structures like in England; we would not survive. In fact, sweating in over-heated living rooms is just as much a feature of winter life as freezing outside. And if you stay active and are well dressed, you don’t feel cold even in severe frost.

    • UVB exposure destroys (photolysis) folic acid. Low gestational folic acid is directly causal to a variety of congenital birth defects (primarily neural tube defects) Those higher birth defects are directly linked to higher infant mortality.
      http://www.ncbi.nlm.nih.gov/pubmed/22852064
      http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2671032/
      So natural selection of human skin pigmentation is a balancing act between maintaining both Vit D production and folate destruction, especially critical in women during pregnancy and lactation.
      Darwin believed that sexual selection for lighter skin mates in humans is the basis for the latitudinal gradient in human skin pimentation. Thus there is the hypothesis that evolution of human skin pigmentation is a competitive balance between sexual selection and natural selection.
      http://www.ncbi.nlm.nih.gov/pubmed/12573076

  12. Willis:
    We do arbitrary grouping all the time in flow cytometry analysis. I’m not endorsing it, just saying.
    Here’s how we do it.— In graph 4, draw a horizontal ines at 13 years old (reproductive cutoff) and 5 years (infancy); now draw a vertical line at 110 sunspots (roughly half the max). The upper left quadrant is far more populated than the upper right quadrant. Therefore, far more young reproductive-competent Norse died when they were born during low sunspot years than died when born during high sunspot years as compared to Norse dying before 5 years under the same sunspot number category. So, it’s the lack of sunspots that’s killing these Hyperboreans when they become disaffected youths.
    Of course, once you factor in the rarity of high sunspot years, the effect probably goes away 8^), not to mention my cherry picking. Is that some reason not to publish? I think we can wring a few more findings out of this data.
    Lovely article, BTW.

  13. Willis as others have noted about the twenty somethings missing there may be a number of reasons why they are not there. I can say as I went through my late teens early twenties with a group of my peers at Church many left for uni or jobs elsewhere so there were fewer of my generation, then a few years later some returned and others arrived as they found jobs in the area so the number in my age group went up again. This wasn’t a small village but an urban/semi rural area outside London.
    James Bull

    • Thanks, James. While the effect you note could cause a decrease around age 20, that’s not what we see in the data. It stops entirely at exactly age 20. That’s the part that is a mystery to me.
      w.

  14. I wonder how likely it was that women in those villages would get married after age 20 or 21. Remember that they lived before modern medicine (and before all but the crudest birth control devices), so there was a strong tendency for women to marry and then die soon after while trying to give birth to their first child. Having survived the first birth, they would then be much more likely to survive a second, third, etc. I had read about this phenomenon, but had no idea it would so strongly manifest itself in the data — if that is what is in fact going on in Figure 1.

  15. While I studied in the 1980-ies I read an Astronomy 101 book for leisure. The headline of this article reminded me of a quote in this book.
    I have spared the book and looked it up, it says:

    “The mean number [of sunspots] has been used and misused in a lot of different areas in attempts to explain, or show a causality on, almost all the phenomenon as can be imagined here on earth”

    ISBN 82-00-06449-2, Elgarhoy, Hauge, Universitetsfolaget,
    Oslo 1983, page 139
    /Jan

  16. Re, the break at 21 years …. National service?
    Wiki says this of the 20th century Norwegian army. Quote:
    “The men were called out at 21, and for the first 12 years belonged to the line ; then for 12 years to the landvarn. Afterwards they passed into the landatorm, in which they remained until they attained the age of 55 years.”
    The quote is not clear, but they seem to start national service aged 21, and so their deaths may be in military records. However, this does not explain the lack of female deaths.
    R

    • I have served in the Norwegian army in the 20th century. I was called in at 19.
      There were no wars in the 19th century that could explain the missing deaths. As previously mentioned, many emigrated to America. Some of them lost their lives in the American Civil War.

      • While mortality might not have been high, there were the Gunboat War and the war against the Swedish invaders, both during the latter Napoleonic Wars.
        The state of Denmark-Norway remained neutral until the Battle of Copenhagen (1807), which helped make Horatio Nelson famous. Denmark was forced to cede Norway to Sweden by the Treaty of Kiel in 1814. After a brief Swedish campaign, Norway entered into personal union with the Swedish crown, which went in 1818 to Bernadotte, one of Napoleon’s marshals, as I’m sure you’re aware.

      • “There were no wars in the 19th century (…)”
        Napoleonic Wars: http://en.wikipedia.org/wiki/Napoleonic_Wars
        Mountain men are naturally tough and hardy*. Bring them down to the lowlands and they are strong and tireless. Swiss pike-men had a fearsome reputation and were highly sought as mercenaries. I daresay Norwegian men could earn good wages as mercenaries, too.
        *Lowlanders are more numerous, because the land is more productive.

    • rallfellis.

      The quote is not clear, but they seem to start national service aged 21, and so their deaths may be in military records. However, this does not explain the lack of female deaths.

      If the death was in a military barracks or on campaign, it will not be in the local church or village records.

      • In a related vein, Parish records would be only for those who were christened and/or buried in the Church’s official cemeteries.
        Church records are not ‘census’ records tracking everyone born, living and deceased.
        Children that died before christening or baptism were usually buried local to the household. As were many children of snow bound families that died during the winter.
        I also wonder about data frequency patterns.
        Data provided by Professor Røskaft is already modified into a summary file; leaving many questions:
        -e.g. Are the min/max records averages over the year or perhaps totals?
        -e.g. Just what does a calendar year have to do with the data? Mankind’s calendar year does not align with solar minimums or maximums. A tally for the year is arbitrary versus an actual solar minimum/maximum period of time.
        -e.g. Were any bell curves plotted?
        -e.g. Given that pregnancies occur over a long period of time, How does the date of a child’s death during a solar max period when their gestation period was primarily during the previous year during a solar min period.
        -e.g. How are records without beginning or an ending treated?
        -e.g. What steps are taken to control for obvious non-solar mortality; e.g. war?
        Statistical significance aside, where are the common sense questions and answers before seeking significance? Many of which, Willis competently demonstrates, are missing.
        -e.g. Is this all of the data?
        -e.g. Is this enough data?
        -e.g. Is this the right data needed to answer all ’cause of death’ questions?
        As with so many other research wasting papers out recently; Professor Røskaft’s paper appears to be one of those with strong impetus to decide results first and then build a pretty data trail to those ‘results’. Only summation data is presented without a trail of work describing analysis progress.

  17. Perhaps those in their 20s and thirties were there, but just not dying, perhaps they were survivors of the previous epidemic.

    • For most epidemics, people in their 20s would be expected to show a higher than average survival rate. In the case of measles, chickenpox or smallpox, this would be especially true if they had been exposed to and survived these diseases as children. But in an isolated rural environment, they might not have. Childhood disease can be lethal to adults. Rural American Civil War (so-called) soldiers died from mumps.
      The numerous Norwegian dairy maids might well have gained immunity from deadly smallpox by their daily contact with cows, vector of the milder virus cowpox.
      Immunization against smallpox was introduced to Norway c. 1801-10. A full-blown public health system was instituted in 1860.

      • So-called refers to the name of the war, not the soldiers. It wasn’t technically a civil war.

    • The 1918-1919 Influenza outbreak was particularly deadly for young adults. The very young and old did not suffer the same mortality rates.

  18. Willis,
    Could you give us an example graph of what it would look like if there WAS a correlation between sunspot numbers and birth. I am presuming that in fig 4 you would get a dark clump pf spots around the 100-150 sunspot section – because it would drag all the long-lifespan spots down into the shorter lifespan regions.
    Thanks

  19. Hi Willis,
    I also graphed the data set a bit and I guess there is conclusive evidence that they had in fact two data sources. If you look at the data, there are no death below 25 years of age for thoise born before 1700 and no death at age above 25 for those born after 1875. The simplest explanation is that in the low age of death data the birth date was for some reason shifted by 25 years – and if this is a systematic error it is only affecting the low age at death data -> thus two data sources.
    I wonder how the authors would explain this shift though.

  20. Willis, I’m wondering whether you could think of a way to improve the presentations of figures like your Figure 3? They don’t satisfy my “inner Tufte” :). Perhaps you make the columns narrower with gaps, or something.

  21. My hypothesis is that there are two people to blame for the exponential rise in junk science. Steve Jobs and Bill Gates. (One has to discount the variables such as Al Gore…the man who invented the internet.)
    With out the availability of computers and statistical programs for dummys a lot of the so called climate scientologists would be out there checking temperatures and gathering useful data. Instead they are in computer labs turning out statistical junk on computer programs that they don’t really understand.
    As someone once said: ” Just because you can, it does not mean you should.”
    It would be good if Willis could find the correlation or causation with this hypothesis. Number of science papers p.a. issued before windows/ number p.a. after windows for example.

  22. Dear Willis!
    I have compared the size of annual age classes for the whole population according to reported age at death in the statistics of cause of death in Finland in 1750-1850 (1,8 mill. people). There is an interesting coincidence at twenty years age. From twenty years upwards there is an uptick in the height of the graph for each age of 20, 30, 40 etc. The reason for this is uncerainty in how old the person was at death. The age was obviously given as “about” 20, 30 40 etc years. Accordingly these numbers are higher than the preceeding or subsequent age classes. This fact must be considered in the statistics when comparing various diseases. I wonder if this also could have affected the present study.
    Larry Huldén
    Finnish Museum of Natural History

  23. “The men were called out at 21, and for the first 12 years belonged to the line ; then for 12 years to the landvarn. Afterwards they passed into the landatorm, in which they remained until they attained the age of 55 years.”
    “Line” is the regular army
    “Landvarn” is the home guard, which cannot be deployed beyond the national borders.
    “Landatorm” is the reserves and used only in time of war for local defense and garrison.

    • PLS

      “The men were called out at 21, and for the first 12 years belonged to the line ; then for 12 years to the landvarn. Afterwards they passed into the landatorm, in which they remained until they attained the age of 55 years.”

      Just in case anybody thinks today’s military is taking a high percentage of your yearly wages …

  24. Coronary heart disease and geomagnetic activity
    http://meetingorganizer.copernicus.org/EGU2014/EGU2014-13457.pdf
    “….During 14 years we collected more than 25000 cases of acute myocardial infarction and brain stroke at seven medical hospitals located in Russia and some other countries. We used only cases with established date of acute attack of diseases. Undated cases were excluded from the analysis. Average numbers of patients at geomagnetic active days and days with quiet geomagnetic condition were compared. It was shown statistically that during geomagnetic disturbances the frequency of myocardial infarction and brain stroke cases increased on the average by a factor of two in comparison with quiet geomagnetic conditions….”

  25. Just single look at Briggs’ blog made the problem with this reanalysis obvious:
    You only work with deaths. To analyze mortality, you need to divide number of deaths (under certain age) by total number of births. Without that, the noise that natural changes in birth rate introduce to the data is too much to obscure any potential results. Of course, you didn’t have total number of births for each year available (though I find it plausible that it was available to the paper’s authors).
    What you have shown is that data do not suggest that sunspots introduce statistifically significant change in mortality slope under 21 years. However, change in mortality slope is order of magnitude harder to detect than change in total mortality.

    • On further investigation of the data I found one oddity. First I limited the data to only 1760-1830 years because these don’t suffer from ‘end of interval’ conditions. When I then created a scatterplot of number of people who survived 21 years versus number of total recorded births to that year, the dependency was linear but did not go through zero. The more people were born in a year, the greater was the chance of their death being recorded. It could be used to estimate number of unrecorded deaths after age 21.
      Apart of that, I found no signs of 11-year periodicity. None at all however hard I tried.

  26. I think the poet was Pam Ayres, who said
    Thump’n shake the plum sauce bot’l
    None’ll come, ‘n then the lot’l.
    Or something like that. Ahead of her time if opining on climate data analysis.

  27. The paper may be confusing cause and effect. [Child] mortality rates will dictate early death, very high before child childbirth and infant illness were understood let alone treated correctly. These figures will be distorted by birth rates, which normally increase. You also have to consider the fact that better living conditions mean that people live longer. Poverty is also a great driver to early death, back when the figures were extracted it was dreadful. Climate also adds to the mix, deaths increase as it cools.
    I do not think that sunspots have much to do with it apart from low spot numbers mean cooler temperatures.

  28. Willis,
    I think there is a time delay of twenty years in your data set because the original recruits to the study are from marriage records. The legal minimum age of marriage is 20. Therefore all deaths from age zero to twenty are recorded as being for children born to married women already in the study.
    Try adding a twenty year static to the age group 0 to 20 and re-plot your graph.
    Some general thoughts on this:-
    1. The study starts with marriage records. The data are recorded by year, not by month or day. This time band discretisation is equivalent to a low pass filter.
    2. What is the age at which marriage becomes legally possible? The minimum age of marriage is not stated but we can infer that it is 20 years from comments about recruitment into the study. If we assume that the society was traditional and that the typical gestation time of 9 months applies, then women aged 20 will only have 3 months in which to give birth before they become 21 and are recorded in the next age band. “Anticipation time” and shot-gun weddings at “showing time” may add another 3 months to this age group, but there will still be fewer births to women aged 20.
    3. Dead singles, of whatever age, don’t get married so they are not recruited into the study.
    4. I don’t know how the data will be affected by teenage mothers who have to wait until they are twenty to be legally married.
    (Memo to self. Light blue touch paper and retire.)

  29. Willis : I have downloaded the data, and there appears to be a pattern related to the position of the birth year in the solar cycle. Surprisingly, it’s of lower numbers of early deaths with birth year at solar min or max but higher in between (on the up and down slopes). Surprising, so I’ve probably got it wrong! It’s late here in Oz, so I’ll come back to it tomorrow, but with your superior expertise (and R) you might like to take a look at how deaths before ages up to say 5, 10, 15, 20 relate to the birth year’s position in the solar cycle. ie, looking at the solar cycle instead of SSN. Wiki gives the solar cycles here https://en.wikipedia.org/wiki/List_of_solar_cycles I was using the start and maximum dates to determine the phases of the cycles.

    • Well, I’ve been back through the data, and fixed a few things. As expected, no significant solar cycle pattern emerged.
      I downloaded the Norwegian data using the link https://dl.dropboxusercontent.com/u/96723180/Norwegian%20Lifespan%20and%20birthyear.txt provided by Willis. I downloaded solar cycle (SC) data from http://en.wikipedia.org/wiki/List_of_solar_cycles
      I divided each solar cycle into ten deciles, five equal periods from cycle start date to cycle maximum date and five equal periods from then to next cycle’s start date. I numbered the ten deciles 0 to 9. So deciles 5 and 6, for example, are just before and just after solar maximum. I think better dates are possible, but this uses publicly available data, so I’m not adding bias.
      The 252 years 1756 to 2007 are not evenly distributed between the SC deciles, partly because the second half of each SC is on average 6.8 yrs vs first half average 4.2 yrs. The distribution over SCs 1-13 (the bulk of the Norwegian data) is similar.
      After allowing for the uneven distribution of the years, there was no obvious SC-related mortality pattern, well not that I could see, anyway. Maybe there are slightly more births just before solar maximum – but that’s pretty much the case across all lifespans, so it makes no sense unless the approach of solar maximum makes conception more likely!
      http://members.iinet.net.au/~jonas1@westnet.com.au/NorwayBirths.JPG
      http://members.iinet.net.au/~jonas1@westnet.com.au/NorwayDeaths.JPG
      (The blue line that stands out a bit in both graphs is the 21-30 lifespan. No idea why.)

  30. @Philip Mulholland
    I think you are probably right BUT we don’t really have any raw data (I mean real, raw data) from the marriage records. The download TXT is a joke. They should be ashamed if this is what they are working with.
    I cannot imagine why anyone would start with marriage records, except that the base data used by the present study comes from a bunch of sociologists. As you say, Philip, you miss a lot of people. If anything you have to start with death records (counterintuitive, I know).
    If only it were so easy. Unfortunately it is a very messy task trying to associate marriage, birth and death records with unique individuals.
    1- Let’s note first that parish records (certainly in central Europe) are not records of births but of baptisms.
    Big deal, you may say. But stillborn and very young infants, if they can’t hang in for the day or so it may take to get them baptised, will not appear in the baptisms list at all (as Philip points out). They will, however, appear in the death register with an age of 0 or some hours.
    In the 18th and 19th the quantity of such ‘lost infants’ was a very high proportion of births. In the area and period that I am familiar with (central Europe, 18c) a family may experience 10 to 15 births, of which usually only a small proportion would survive to adulthood and where a number of those births would usually be ‘lost infants’.
    2- People move around. The baptismal record does not tell you when and where that person died. The death record usually tells you where they are ‘from’ which can also be guesswork.
    3- The age on the death certificate – particularly if the death does not take place in the parish holding the baptismal register is often just a guess, and can be a wild one, too.
    4- Some people marry, some don’t, some marry a lot (common in the 18c).
    For the above reasons it is very difficult and time-consuming to try to come up with a list of unique individuals. The sociologists in the underlying study were just interested in socieconomic groups and not absolute births and deaths.
    Ahh, Sociology, the queen of sciences!
    It would have been nice to first plot a chart of age against frequency. That would make it much easier to decide if the data has superficial problems. The age at which a particular person dies is not just determined by genes and dice-tossing. Each age has characteristic dangers. We could then see easily whether the data points for age correspond to reasonable values in terms of our expecations:
    We expect
    1- a big peak at 0-1: infant death is a dramatically high proportion.
    2- a peak at 1-5: during this time the immune system gets tested and built up. Many fail.
    3- that once a person has got to about 10 years old there will be good life chances until about 40.
    4a- that if a person reaches 40 intact they have a good chance of achieving an age between 60 and 80.
    4b- special case: women. Women will come into childbearing age about 15 and may not survive the procreational load. Death during childbirth tends to be earlier rather than later. It’s the first or second child that gets you. After you have had 15 kids you can be considered indestructable. This may explain the suspicious looking straight-line cut-off before 20 y/o.
    5- there will be outliers at the top. Apart from genes and chance there is no pattern here: I have records of 18th century people who have had extremely hard lives but have nevertheless lived to 80+ (in one case I have, 101 years).
    Given that the sample is ‘only’ 5k+ and is distributed over 200 years, there aren’t really a lot of code points in each year, once we take away the heavy loss caused by infant mortality. This also may explain the very empty 20-30 range.
    We not only have to allow for epidemics, we also have to allow for weather disasters. For example, the ‘hunger years’ in central Europe of 1769-73 (the endpoints a bit vague, depending how you measure them) went like a scythe through the population. The were caused by a a succession of very late, very cold and very wet springs.
    Now, you WUWT guys will certainly be able to tell me whether that had anything to do with sunspots. Come on, get your spreadsheets out!

    • That’s a grea post, Peter. But by “Sociology, the queen of sciences!”, you really mean: “Sociology, the Fairy Queen of all sciences”, don’t you ? 🙂

  31. It really is like two different data sets. I loaded into LibreOffice Calc and sorted by year. Starting in the 1800’s things seemed to change. Starting at 1866, or so, they record very few life spans greater than 20, and a very high percentage of 0s. Strange.

  32. There are many potential problens with this data, but it also reflects what must have been an extreme infant mortality. In a nation where extreme poverty was the norm, farming is difficult and malnutrition and undernourishment would have been the norm, that is hardly surprise.
    but inferring mortality from church books is flawed, as it gives burials not deaths. In a nation where going off fishing was a major source of income for much of the population for parts of the year, the books would not have accounted for those many who died at sea, a fairly large number one imagines. Sea faring in general being another major industry, those who died abroad would likewise be unaccounted for.
    But its not necessarily that wrong, I imagine people in their 20s were less likely to die than most other ages, being at their prime and having survived the danger years and the various diseases they’d contract naturally as children.
    And considering the gaggle of children people had at the time (around 10 seemingly the norm assuming a healthy couple, and over that not uncommon) death in childbirth seems surprisingly uncommon. If any group in their late teens and 20s were to be represented in the books, the women are the ones most likely to be present from cradle, or atleast marriage, to grave. Perhaps the same study excluding males would give a more representative group.
    With nearly 100 % of the 1825 equivalent population leaving the country by 1925, the data is severly deficient.

  33. I looked at the original paper and this is all quality stuff …. for the Onion! All seriousness and earnest, but absurd. Who looks for UV effects in a country more famous for lack of sunshine? Near the arctic circle no less. There is nothing in this paper to make me believe that the subjects ever spent much time outdoors. Maybe the inland subjects were farmers and the island subjects were fishermen, but who knows?
    Every once in a while, someone should do a study like this just to show that statistical techniques should never be used by those who do not know their limitations. The limitations of statistics, but also of sociologists evidently. So many factors, so few controls.
    The paper would have been better if it referred to the subjects’ signs of the zodiac. Really!

  34. Hi Willis , its possible that the effects of volcanic activity from nearby Iceland could have messed with their study findings by blocking the sun , crop failure etc. Perhaps taking icelandic erruptions into account may even improve the corralations they where looking for ?. There is a timeline of eruptions at a site called ; ICELAND GEOLOGY ‘ a short history of volcano eruptions in iceland ‘ . There was rather a lot of them back in the times of the populations studied. cheers

  35. I can understand the infant mortality rates, look at the graves in any cemetery that covers the pre-modern medicine & obstetrics era & you’ll see a horrendous number of children’s graves.
    As for the rest, I’d expect there to be a low level of mortality amongst young adults, especially men. Once they’ve survived the diseases of childhood, child-birth aside (hence the men!), they’d have a good chance of living a full life. I am surprised by how old they lived to. Looks like around 80!
    The cut-off at 20-ish is odd.
    Over to the medical profession!

  36. Controlled maternal effects?
    That would have to be the biggest factor. From Wikipedia (a bit lazy today)
    Young mothers face higher risks of complications and death during pregnancy than older mothers,especially adolescents aged 15 years or younger.

  37. The first rule when looking for correlation between two variables is there must be a plausible causal link between the variables. If not, it doesn’t really matter if the correlation is statistically significant. Rare events do happen by chance. It’s not that difficult to get two pair in a poker game. The odds is one in twenty. That would be statistically significant for 95% confidence interval.

  38. Correlation studies comparing health to some third factor are really difficult because health effects are dependent on condition and also dependent on the presence of disease which may or may not be related to the third factor. So, for example, you may have been poor condition in 2009 but as it turned out to be a low flu year even in countries that didn’t vaccinate the chances are you survived (and the fact that H1N1 wasn’t very virulent also helped- virulence of a disease strain is also difficult to distinguish from environmental effects) . You need to determine if epidemic years correlate with sunspot actvity and if they don’t you still need to check if in epidemic years only, does sunspot activity correlate with, for example, mortality for epidemic years. This subdivision of the dataset greatly reduces the number of cases in the analysis and typically lowers the power of the trial so you have to lower the criteria for significance to avoid a type two error. A proper evaluation is very difficult and in the end it is only correlation so one has to ask if it is worthwhile. At this point I would have to say that Willis’s analysis makes it seem likely that sunspot activity doesn’t affect the likelihood of epidemic but it doesn’t help us decide if epidemics are worse when they occur in high or low sunspot years, so the question has not been fully addressed

  39. I’m confident the authors know correlation does not imply causation.
    I’m also confident the authors know correlation helps insure publication.
    Any correlation.

  40. “Solar activity at birth predicted infant survival and women’s fertility in historical Norway”
    I didn’t know the Royal Society publishes astrology. Oh well it also publishes AGW propaganda.

  41. Regarding the difference between British (or English if you prefer) and Norwegian houses, the climate is different. In the part of England where most of the population of the UK actually live, snow is a rarity, at most a few days per year. Any house built before 1966 was built with a fireplace for a coal fire, and what now counts as draughty was then ‘well ventilated’, as an open fire is more likely to kill you in a sealed building than cold in a draughty one. A single coal fire in one room of a poorly insulated traditional British house heats the whole structure to a tolerable or better level in a British winter, and when the climate was that bit colder in the mid 19th C, more than one fireplace was provided, up to one per room, and for all I know up in the North. But there the diet is heavy on fried Mars bars, pizza fried in batter and Haggis, washed down with IrnBru and alcohol, so the inhabitants themselves are physiologically more hardy.
    Following the decline of coal as a primary source for domestic heating (blame who you like: Clean Air Acts, Mrs Thatcher, the Aberfan disaster or just getting fed up of the dirt), British building standards have seen a progressive change. It would now be unusual to see a new building without cavity wall insulation, double glazing, central heating etc. with the central heating run off gas, electricity or oil, and in transitional cases coal. (And coal provides most electricity too).
    I could live in an unheated tent in southern Britain year round and through the winter, but I wouldn’t like to even try it in Norway except in High Summer, and even then I’d be bothered by the flies.
    Our lack of snow brings one additional benefit: we don’t have that millefeuille of snow and dog excrement that you get in Scandinavian cities that make the places so obnoxious when it melts. Not that I like being splashed by a passing vehicle here, but it is hyper unpleasant where it happens routinely and one is deluged in liquid manure.
    As for the original thesis, no doubt the statistics in it are twaddle, but infant and childbirth mortality in the past was radically different than today. Indeed, prior to antibiotics, people died of tooth extraction and infections of no consequence today and many formally fatal infant complaints are things of the past.
    Are Norwegians terrified of a degree or two rise in average temperatures? I bet it’s still bloody freezing in a Norwegian winter!

Comments are closed.