Friday, 18 March 2016

New Analysis of Nessie and Sea Serpent Reports

Charles Paxton has written on the statistics of aquatic cryptids before and has had a new paper published in the latest issue of the Journal of Scientific Exploration (Volume 30, Number 1). You have to become a paid up member of the Society for Scientific Exploration to access the article or wait two years for it to become free. I did get a copy from Charles having helped him on this article, so can make a few points.

The article is entitled Consistency in Eyewitness Reports of Aquatic “Monsters” and is co-authored with Adrian Shine. With Charles' permission, I include the abstract for the article below:

Little work has been undertaken on the consistency/repeatability of reports of natural historical  anomalies. Such information is useful in understanding the reporting process associated with such accounts and distinguishing any underlying biological signal. Here we used intraclass correlation as a measure of consistency in descriptions of a variety of quantitative features from a large collection of firsthand accounts of apparently unknown aquatic animals (hereafter “monsters”) in each of two different cases. In the first case, same observer, same encounter (sose), the correlation was estimated from two different accounts of the same event from the same witness. In the second case, the correlation was between two different observers of the same event (dose). Overall, levels of consistency were surprisingly high, with length of monster, distance of monster to the witness, and duration of encounter varying between 0.63 and 1. Interestingly, there was no evidence that sose accounts generally had higher consistency than dose accounts.

If you don't read the article, the one thing to take away from the abstract is the consistency between estimates of distance and size between multiple witnesses and multiple witness accounts. That, of course, does not equate to accuracy, but neither does it exclude it. Though there will be caveats, this can be taken as a positive for witness integrity rather than a negative. I say this because a low precision would not be consistent with high accuracy.

Charles' study used a dataset which was a combination of sea and lake monster reports. It would be interesting to see further results for the single case of the Loch Ness Monster. Note that this does not address the other issue of witness descriptions of what they see, which is a more complex affair than estimating a simple number.

On the cryptozoological theme, I note the same issue carries a review of a new by Karl Shuker entitled: "A Manifestation of Monsters: Examining the (Un)usual Suspects" and it is reviewed by a man well known to Nessie researchers, Henry Bauer.

The author can be contacted at


  1. This is very interesting and raises some questions:

    Are the correlations the same for reports that were later shown to be hoaxes or mistakes, as for those that remain mysterious?

    Can these correlations be explained by commonsense factors? A large animal seen at short range for a long period is not likely to remain unidentified, and is therefore not likely to generate a monster report. This effect will be larger if the witness is an experienced wildlife watcher.

    1. Not sure, would the sample of such cases be large enough to be statistically significant?

  2. The end game, as to Nessie, will not be solved with photo/video (unless out of water) It will be a combination using sonar and DNA samples taken by means acceptable under Scottish law and be accepted as valid by the Natural History Museum.