Iirc, Andy did some experimenting along this line several years ago and got a strong correlation between the refractometer readings and the baked puck data. So the refractometer readings, properly scaled, can be used as a measurment of extraction. However, there wasn't enough of a data set to determine whether the correlation was unaltered when using different coffees, baskets, grinders, etc. Worst case, the refractometer readings need to be scaled differently under different conditions.

I'm not sure how much data Vince collected, and whether a peer review would pass on his scaling algorithm. When I first heard about it, I was personally loath to take on the job of collecting a dataset large enough to create a sound conversion table (for instance, I doubt the SCAA did enough in the 50s for their conversion from conductance to yield to be sound, their raw data never saw the light of day, nor did it make it into a peer reviewed article)

On the other hand, in a shop environment, a conversion table is not necessary. Suppose the refractometer (or conductance) reading is monotonic with the solid + solubles concentration in the cup. That is, reading = function(solids + solubles) and the function is invertible. Let the inverse be S(*) the unknown scaling function that converts the reading back to the concentration. This scale is monotonic, and can be treated like the actual measurement in arbitrary and unknown physical units. Like units of old, they can be different from coffee to coffee and shop to shop, but within a tight context, they work.

Basically, one can, for every setup (grinder, blend, baskets), take refractometer readings and divide them by shot weight to get this arbitrary but accurate extraction measure. One can therefore reliably use it to learn how that measure changes when changing dose, shot time, etc., and which extraction level tastes best. One can then spot check that the same extraction value is being maintained in operation. This is no different than what one would do if the scale was correct.

So I think Vince's system can work whether the scale is correct and transferable or not, as long as it is consistent in any one setting. On the other hand, the same can be said for a $30 brix meter on its own. The software also has convenience and administrative tools; and for a cafe that may make it very worthwhile. I'm less sure if these extras' are valuable in a home setting.

Personally, since there are many brix meters available with a wide range of costs and features, I would prefer if we used raw brix and brix*shot_weight/dose_weight readings to communicate concentration and the extraction measures. That way we can compare notes, and discuss possible conversion algorithms in the open context of scientific discussion. If we use the Terroir system while it is unpublished, we are using something that seems very odd to me, a physical unit which is proprietary and trade secreted.

On "brix meter" refractometers, the most common kind, the refraction angle from the refractometer is converted into a brix reading, which is the amount of dissolved sugar in the liquid. The amount of dissolved sugar will be directly proportional to the amount of total solids that get into the cup for a given coffee and brewing method. Whether it is the same proportion for all coffees and brewing methods remains to be seen. Since the amount of sugars in coffee varies by varietal, region and degree of roast, I doubt it.

Andy, you're absolutely right on the refractometer and the units, but I'm not ready to concede my main two points, which are:

-- The consistency of correlation of refractometer readings to solids concentration and extraction is not known. Your data are a start, but it'll take data from a variety of settings to get some certainty. Pure noise is less of a problem, since that can be dealt with by doing multiple readings (good practice in any case, since there's always operator error, and a huge plus for a simple, non-destructive method like the refractometer over puck baking). I'm more concerned with systemtic errors due to different conversion constants for different coffees, grinders, baskets, etc.

-- The conversion that Terroir uses is unpublished. Until it is, it would be better to communicate with the most widely available units, e.g. brix for refractometers and CaCO3 concentration for conductivity meters.

These points are temporary. More testing will reveal the method's limit conditions, and a patent application will state the conversion algorithm.

BTW, while algorithms are patentable, a patented conversion algorithm may not be much protection. Anyone who has college algebra can do a Taylor expansion of the algorithm for the numerical range of interest, then use any equation that has the same expansion for that region as the "new improved" conversion algorithm.