I was able to find at least one paper in which a standardized wort was used to classify many (153) yeast strains based on, among four other parameters, degree of attenuation.
The wort used was an
'All-malt hopped wort (specific gravity 1040, pH 5.0, attenuation limit ca. 1006) [...] prepared according to [another paper]...'
This second paper referenced indicates that Fuggle hops were used at a rate of one pound per barrel, with resulting beers having ~20 BUs (ranging from ~15-25, and it should be noted that changes in BUs had no effect on the wort fermentability). Distilled water with 200 ppm CaSO4 added was used to prepare the wort.
So, it seems your intuition was pretty much right, at least in the case of this one study (just the proportion of unfermentable sugars is slightly higher, at 15%). I can't speak to whether any given yeast laboratory uses something like this as their standard (or whether they even use a single standardized wort, or rather collect data from performance in a large variety of worts), but this at least gives a precedent for use of a generalized wort to assess a given yeast strain's attenuation characteristics.
There are a few things I'd like to note, as well:
- The first study I linked to rates yeast's attenuation on a scale from 1-5, not in the more familiar % range, to allow for ease in characterizing a large number of yeast strains according to a broad range of parameters (153 strains x 5 parameters = 765 data points). I can't see any reason the methods shouldn't generalize to determining the degree of attenuation as a percentage.
- The same paper gives attenuation data for 20 consecutive trial fermentations using one of the 153 yeast strains tested, showing the average final gravity to be 11.1 (or 1.0111), with a standard deviation of +/- 0.16 (1.0016). This equates to an average attenuation of 72.3% (100% x [40 - 11.1] / 40). Since ~99.7% of data points from a normal distribution should fall within +/- 3 standard deviations of the mean, we can calculate that the attenuation of this strain should be between 71-73.5% nearly 100% of the time. This seems to coincide well with the ~2-4% attenuation range you find on virtually all yeast strain data sheets.
- It's good to keep in mind that 'official', in the sense you use it, doesn't really apply here because (as far as I know) there is no compendium method (ASBC, EBC, IoB etc.) for making a wort used for determination of attenuation. If you're talking about one particular yeast laboratory, their methods are more than likely proprietary, and probably use wort that is made in-house (i.e. White Labs SD has a brew-house producing for their taproom, and I'd guess this is where the wort comes from). That is to say, the official-ity of a given method really only applies within the company.
- I think you're exactly right to want to know the composition of the wort used to make these determinations, because knowing this allows you to guess at the limitations of the information testing these conditions gives you. It's kind of like a Congress mash: many important malt parameters are defined by this method, but many people don't know that the Congress wort is only ~8 degrees Plato, meaning you can't necessarily generalize characteristics like color and pH to a more concentrated wort. However, I think this is just another part of brewing where the manufacturer can only offer a range of expected performance (you don't see yeast companies guaranteeing the attenuation of their yeast!) and so it really comes down to the individual brewer being willing to familiarize him/herself with their particular combination of yeast, wort and fermentation process to determine the best-performing strain (though the numbers provided are generally a good starting point).
- Remember, also, that laboratory results do not always correlate perfectly to real-world performance. The first study uses EBC 2 liter tall-tube fermenters for trial fermentations, which are generally regarded to give similar performance to larger fermenters. However, take as an example a finding from the first paper linked to in Mr_road's answer (which also use EBC tall-tubes): 'Under laboratory conditions, [strain one] seemed most suitable for high-gravity brewing, as it showed the best final attenuation of all the strains. However, in pilot-scale conditions [...] (using completely different parameters), [strain two] seemed to be the best variant.' If yeast attenuation ratings really are determined as a series of trials on a standardized wort in a laboratory setting, it's worth noting that they were probably conducted in a similar fashion and the results obtained would not necessarily apply exactly to larger-scale fermentations, nor to every possible wort.
Anyway, this answer ended up taking way more time and effort than it probably should have (...), so I hope it helps.
I don't have access to the EBC Methods of Analysis, which I thought might specify a standardized wort for the tall-tube fermentation method I linked above. I do, however, have the Institute of Brewing's Methods of Analysis which, although no longer in publication, have as of 2005 been 'fully integrated with the Methods of the EBC'. It contains the method 21.25 Yeast Characterization: Fermentation in EBC Tubes, which is presumably comparable to (if not the same as) the EBC method. Unfortunately, it specifies only that 'hopped wort' is to be used, giving no further detail on its composition.
I was able to check out the ASBC's miniature fermentation assay method, which is used 'to compare the fermentability of yeast strains'. The wort used is actually prepared differently than I expected:
- First, wort is prepared using a two-step mashing procedure (45°C to 70°C). The malt type is not specified (this method is actually also used for assessing malt performance). Wort is diluted to 12.6°P (1051 SG).
- Dextrose (glucose) is then added to achieve a gravity of 16.1°P (1066 SG).
- Wort is aerated 'at 20°C for 5 minutes with medical-grade, compressed oxygen'.
- Yeast is pitched at 15 x 106 cells/mL and fermentation takes place at 21°C.
So you can see this is quite different from the first paper I referenced. Note, especially, that this wort is neither boiled nor hopped. This might be the standard to which White Labs was referring, but there's no way of knowing for sure.