It has been a while since I last vented about the continued use of the so-called “Tangent-by-tangent” (TBT) method. This method ostensibly is able to determine accurate size distributions using nothing but a superposition of Guinier functions. The last time that I talked about this (probably) was in the pitfalls-section of the Everything SAXS review paper. But now there’s good news!
A little bit of background. The TBT method first appeared in 1946 in a paper by Jellinek et al.. They indicated that it was a distinct approximation that looked useful, upon which it was correctly noted: “At this stage of development it would be premature to take too seriously these computations of particle size distribution.”
If you take a look at what happens to the Guinier method for polydisperse systems, you see that it assumes the volume-squared weighted mean radius of gyration, as re-demonstrated here and here. The idea that you can pick apart a scattering curve originating from a monomodal size distribution of objects, by using a superposition of Guinier functions alone is, therefore, farfetched.
Despite these demonstrations, the TBT method is still in use today. Currently, it is still being used by a group in China, who also wrote some software based on it (and whom I personally confronted at SAS2012). There is a steady drip of seemingly interesting papers culminating from their efforts which casually mention the TBT method. One of the latest papers claiming its successful application is this one.
In order to prevent other researchers from wasting their time on this (risking discrediting the field in the process), it is useful to let people know that it doesn’t work. I do so from time to time, but I was happy to see an actual paper a few weeks ago which does just that. The paper appeared in a Russian institute journal which is not so well known and may not have a large readership, however, with some amplification it may turn out to be a useful tool in the battle against bad SAXS.
Have a read, let me know what you think below!