Who estimated war casualties from tightly-controlled government news sources? I've read about this historical case before so I thought it would be very easy to Google, but after a few dozen queries that come up with nothing relevant I'm ready to punt.
The story goes like this.  There is a war on  (maybe WWI?) and for strategic reasons the government wants to avoid disseminating useful information about the number of casualties they have sustained.  As such, they don't publish obituaries for everyone who has died in the war.  Instead, they limit the number of obituaries printed each week to some number $n$, where $n$ is far lower than the actual number of weekly casualties.
However, it's noticed that in the obituaries they first see obituaries mainly for so-and-so's first son, and later on the proportion of second sons increases, then third sons, etc.  Making some assumptions, they're able to successfully estimate casualties from this information.
Does anyone have the details?
 A: You may be referring to Hugo Steinhaus. He was a Polish-Jewish mathematician, which resorted to the ingenious method you sketch to estimate the casualties of the Nazis, while in hiding during WWII. What source I got this information from, I cannot remember. It was a book, which I may have given away to someone long ago; but I checked the information from
https://hmong.es/wiki/Hugo_Steinhaus (in Spanish) a moment ago, and you will surely find sources in English if you google for Hugo Steinhaus.
A: Adding to the answer by @F.Tusell: The english wikipedia article on Hugo Steinhaus contains

Also while in hiding, and cut off from reliable news on the course of the war, Steinhaus devised a statistical means of estimating for himself the German casualties at the front based on sporadic obituaries published in the local press. The method relied on the relative frequency with which the obituaries stated that the soldier who died was someone's son, someone's "second son", someone's "third son" and so on.

It also links to: Kac, Mark (1974). "Hugo Steinhaus--A Reminiscence and a Tribute" which contains

I cannot resist giving one more example of Steinhaus’ quick mathematical
intelligence. It has to do with his estimate of the casualties of the German army in 1944, and it should be borne in mind that he was then in hiding and completely cut off from any source of reliable news.
He noticed that some of the obituaries of German soldiers which were published in the rigidly controlled local newssheet mentioned that the dead was the second or even third member of his family to have fallen in the war, and this was information enough!
For by dividing the percentage of obituaries of second, third, etc. sons by the
(conditional) probability that a family with at least one son will have more than one,
an estimate of casualty percentage can be obtained. Disregarding the age factor (some
sons may be too young to be drafted), all one needs is the average number of sons in a
family (easily estimable) and the knowledge that the number of sons obeys the
Poisson distribution.

