Sun Mar 07, 2010 11:08 am
There is a problem in the spreadsheet. The formula used is RA = SRM*12.2 - 122.4 for the lower limit and RA = (SRM-5.2)*12.2 for the upper limit. This implies that color data was plotted against SRM and a linear fit performed. This is a very interesting thing to investigate but there are lots of problems. If you want to plot Guiness, for example, should you look at Guiness Draft which measures 47 SRM or Guiness Extra which measures 67? And what should you use for the RA? Are they both brewed with the same water? Assuming they are which of several analyses for Dublin water available for Dublin should be used considering that 1) most water reports are seriously in error and 2) Guiness isn't brewed at St. James Gate any more. Taking the approach of using both beers and all three available water reports and doing similar things for a handful of other beers I get a scatter plot which isn't terribly confidence inspiring in terms of doing a tight fit to it. Indeed, the best linear fit has Pearson's R (remember that Gosset, known as Student, worked at Guiness and studied under Pearson) is less than half (0.46) which means a very poor fit. Nevertheless, the slope is more like 1 than 12 so that RA = 5 + SRM better represents the data I looked at that either of the formulas above. For 32 SRM this new formula would give an RA of 37 with the 95% probability band spanning 11 to 67 ppm as CaCO3. An RA in this range is certainly more reasonable than 268 - 327.
So my recommentation for this stout would be to brew it with your lovely (assuming they got all the mercury out) water and check the mash pH. If it comes in too low (unlikely) then add some chalk to the mash. If it comes out too high (more likely but not very if you have 10% or more roast malt in the grist) add some acid to the mash.
I really do hope the Palmer spreadsheet gets fixed as it is used by almost everybody many of whom are led astray over this issue.
2011-2012
