A while back, there was some buzz about a study showing that, to quote the media reports, "Finger length predicts physically aggressive personalities". Like everyone else, I wondered what my finger length said about me.

You can get a pdf from here. The authors found that mean index finger:ring finger ratios were 0.947 (M) and 0.965 (F). Here's their method:

Scanning was conducted prior to examining or analyzing questionnaire scores. A Hewlett Packard Scan-jet 5400C was used to scan participants' hands. Before scanning, small marks were drawn on the basal creases of the index and ring fingers using a ballpoint pen by the first author. This was done to increase accuracy because it was difficult to see the creases clearly on the scans. Both of the participants' hands were scanned at the same time, palms down. Participants' index (2D) and ring (4D) fingers were measured from the hand scans using the GNU Image Manipulation Program (GIMP). The total length of each digit in units of pixels, from the middle of the basal crease to the tip of the finger, was determined using the GIMP "measure" tool. The first author took all of the measurements. Ratios were calculated by dividing the length, in pixels, of the second digit (index finger) by the length, in pixels, of the fourth digit (ring finger) for both hands. This technique provides good reliability (r = 0.98, d.f. = 8, P <0.01 blind test-retest of 10 individuals each scanned twice, with one week between the two scans).

I found that it wasn't at all difficult to see the creases on a scan (Epson Stylus CX7800, 300dpi), but choosing which crease to call the baseline is not entirely straightforward. I only scanned my right hand, as the authors found stronger sexual dimorphism on the right than on the left hand, and this is consistent with earlier literature:

Here's a closeup of the base of the fingers (ring on the left, index on the right):

See what I mean? Even if you draw a line with a pen, where do you draw it? You have to decide by eye: if you try folding the fingers towards the palm in an attempt to use the fold to direct the pen tip in some sort of objective manner, the skin is too loose to get a consistent result. Next, I drew lines on the crease closest my palm using the line tool in Photoshop, and delineated the end of each finger using the freehand lasso tool to identify the far edge:

So as to be readily visible on the web, that image shows a 2-pixel line, thinner than you could get with most pens, but for the actual measurements I used a 1-pixel line. I rotated the image until the finger axis was as nearly horizontal and the crease as nearly vertical as possible, then cropped from crease to end of finger; according to this method my ring finger is 885 pixels long and my index finger 902 pixels, giving me a ring:index an index:ring ratio of 0.981 1.02, higher than even the female average.

Or is it? The averages I quoted are from just one study, and even my brief attempt at a home-made replication shows that there could be significant measurement issues here. Further, whether or not my measurements and those averages are accurate, what does it all mean? How strongly does this particular morphological measurement correlate with, say, aggression; what's the proposed mechanism behind the correlation, and what other correlations might that predict?

Tune in next time (it's Sunday and I've been on the damn computer all day!) to watch one scientist (me) apply general principles of scientific reasoning to questions outside his own experience but (I hope) within his competence.