Hacker News Reader

Comment by gwern | Hacker News Reader

The paper in question: "ON AN ANCIENT EGYPTIAN METHOD OF DIAGNOSING PREGNANCY AND DETERMINING FOETAL SEX", Ghalioungui et al 1963 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1034829/

I'm mostly surprised that they cite two previous experiments by Manger (1933, "Untersuchungen zum Problem der Geschlects diagnose aus Schwangerenhamn" https://www.gwern.net/docs/biology/1933-manger.pdf ) and Hoffmann (1934, "Versuche zur Schwangerschaftsdiagnose aus dem Harn" https://www.gwern.net/docs/biology/1934-hoffmann.pdf ). Who knew? Truly an example of Cowen's second law.

View on HN

The pregnancy testing has been tested on livestock (opposite result, pregnancy inhibits growth)

https://www.researchgate.net/publication/280624411_Germinati...

http://www.veterinaryworld.org/Vol.3/March/Seed%20Germinatio...

The Magician and Leech by Warren R Dawson doesn't seem to be online. Which makes it seem like a pretty cool book.

I see no proof their was an ancient Egyptian pregnancy test, all articles seem to revolve around the one story. Seems like a back history story.

Hieratic Papyri from Kahun and Gurob is online (page 11 onwards for the gynaecological scrolls, it's interesting)-

http://www.etana.org/sites/default/files/coretexts/15145.pdf

The main results of the experiment in the first paper is:

They used 48 samples

* 2 male -> no grow

* 6 non-pregnant females -> no grow

* 40 pregnant females -> 12 no grow | 5 poor grow | 23 grow

Mi guess is that in the press article they calculate 28/40 = 70%.

So no false positives? That’s probably why they kept on using it - if it grows, you can be sure that you’re pregnant.

I can imagine it might be easy to optimize this test a bit...

By repeating the test every few days, and looking at the patterns of grow/no-grow you would probably get a much better result.

With only 6 (8?) cases, the naive bound of the false positives is ~15% and if there is some statistician nearby the bound will be much higher. (I guess close to ~25%, but I'm too lazy to look up the correct calculation.)

And if it doesn’t grow you can repeat the test in a week or two - even modern tests have this issue since early in the pregnancy hormone levels might not be high enough yet.

Regarding false positives - since you’re detecting hormones that are not produced outside of pregnancy time, then the only possible reason for false positive could be some genetic condition, or maybe some serious endocrinologic disease, maybe ovarian cancer.

Reply on news.ycombinator.com