As I suggested a couple missives ago, the controversy around the work done back in 1960 on Thayer's Gull deserved further investigation. So, sparing no expense, I've managed to obtain a copy of N.G.Smith's original monograph on his work with gulls in the great white north. I've also secured the paper's written by Snell calling his work into question, Smith's response, Snell's response to Smith response, etc. One doesn't get the impression that these two men like each other.
After a careful reading I've come to a conclusion. Both were right if one factors in context. Most of the issues surrounding Smith's work can (in my opinion) be explained without invoking fraud or even sloppy science. I think much of what happened can be chalked up to youthful exuberance, pre-conceived expectations, confused navigation and time. The charge Smith's work has failed to be replicated is not nearly as damning as it would be if it had been carried out in a lab. Field based ecological experiments are far more sensitive to time, and external conditions that cannot be controlled. One gets the impression that those who tried to replicate Smith's work may have been hampered by their own pre-conceived expectations that what Smith did was impossible. I haven't read every paper on every attempt, but that seems to be an underlying theme in replication efforts. Maybe the era of scientists being able to acquire adequate time and funding for field work like this is gone. Maybe scientists are human and see what they want to see.
Smith makes it very clear in his original paper and in his response to Snell's that he went to Home Bay expecting to find a mechanism for positive assortment between similar looking gulls of different species and that it would probably be something subtle like orbital ring color and/or mantle and primary tip colors. He was also operating from an initial assumption that the gulls he was studying were different species and that the differences between them were diagnosable.
By my math, Smith would have been about 23 years old in 1960 carrying out a complicated experiment in a complicated environment pretty much by himself. He had really bad maps of the area. His documentation of where he was and what he was doing from day to day were, at least in hindsight, inadequate. It's also not clear how much of a role his assistants played in data collection.
It could also be argued that presenting any conclusions other than those he came to would have been greeted with the same kind of venomous skepticism from the ornithological establishment of 1960 that his work gets today from people who question his conclusions. It's not easy being a grad student.
So, here's what I think happened. I think Smith went where he said he went. I think he and his assistants did what he said he did. I think there may have been more work done by assistants than has been acknowledged. I think the assumptions he made about the diagnosability of Thayer's, Kumlien's and Iceland Gulls were probably faulty, but not particularly surprising given the leaning of the ornithological establishment at that time. I think he saw what he saw and arrived at the conclusions he came to using a filter of expectations not at all uncommon in science. It's easy to label this bad science in hindsight, that doesn't necessarily mean it was bad science in situ.
I also suspect that if someone were to attempt a replication of Smith's work without any filters of expectation that claim the work is impossible, they will find that it holds some truth. Not nearly as much as Smith saw, but more than what Snell is willing to accept.
And it doesn't change the substantial volume of work since 1960 that demonstrates that the Thayer's/Kumlien's/Iceland complex represents a cline of a single species (or perhaps two good species with a big undiagnosable zone of hybridization called "Kumlien's" Gull in between).
Posted by mbalame at January 27, 2005 2:33 AM