The sources of sour
I have recently been running a series of experiments to tease out the relative significance of various factors that contribute to the sourness of sourdough bread.
In the "everybody knows" category is the notion that whole grain breads can be more sour than a 100% white flour formulas, but the relative importance of specific mechanisms by which the sourness is developed is not well documented.
There is a limit to how much acid you can develop in a 100% white flour starter irrespective of how long you let it run before use. This limit is determined by the LAB population and their acid production rate. I have found no references that either differentiate between (or correlate) population growth rate (inverse of doubling time) and acid production rate. While the chemistry of acid production is common to reproduction, it is not clear that the LAB stop producing acid when they stop reproducing. The fact that growth stops at a pH of around 3.8 while there is ample evidence that additional acid is produced down to a pH of below 3.6 feeds a suspicion that at some point there is a decoupling of reproduction from metabolism and the concomitant acid production.
I set as an objective the development of a sourdough loaf that appears visually to be all white flour yet has a reliably reproducible high acidity.
Hamelman has a number of variations on sourdough with some amount of whole grain flour in each. In some, his levain is made from 100% white flour and in others some fraction of the whole grain flour is incorporated into the levain with the remainder incorporated into the final dough. As part of this exercise I began to wonder if there would be value in putting all of the whole grain flour into the levain in an attempt to drive up the LAB population in the final levain (i.e., do not focus on the acid contribution of the levain, but rather focus on how many LAB get added to the final dough). The thought was that a higher numerical density of LAB in the final dough could produce more acid during bulk fermentation and proofing (when the pH is well above 3.8) than they could produce under the limiting conditions of the levain build (relatively small quantity and the development of a low pH well before maturation of the levain). If by adding all of the whole grain flour to the levain could hold the pH up long enough for the LAB to double one more time before they stopped reproducing (relative to an all white levain) then the final dough would have potentially twice as much acid (subject to the other resource and rate limiting conditions of the bulk fermentation and proof).
The results seem to confirm the conjecture but also raise another issue. While including 15% of the total flour as white whole wheat and putting it in the levain makes the bread indistinguishable from an all white loaf in color, it does add acidity to the flavor profile. Yet a much more noticable difference in flavor emerges when this bread is retarded. My speculation is that the relative metabolic rate advantage of the LAB compared to the yeast during retardation (either at 50°F or 42°F) is the source of the additional acid. Some samples of this formulation have produced crumb pH below 4.0 and TTA values of above 12 (ml of 0.1N NaOH to titrate 15g of macerated crumb in 100 ml of distilled water to a final pH of 6.6). Without any whole grain flour or retardation the TTA is 8.70 to 8.85, while with the substitution of 15% of the flour with white whole wheat the TTA climbs to ~9.5 for no retard and to above 11.0 when the dough is retarded in excess of 6 hrs. The sample size for these tests is not yet large enough to determine the significance of the difference, and there are other factors that are not yet firmly under control (e.g., the time/temperature profile for levain growth), but the direction of future tests seems clear.
The next step is to more tightly control the levain, increase the sample size, and add another test case where the whole grain flour is added to the final dough mix and none is included in the levain (with and without retardation).
Since the test matrix is getting too large to explore completely, I will have to make a decision about what cases to emphasize and which cases to defer.
I would be interested in comments from anybody who has managed to read this far without falling asleep about which cases seem most interesting or worth the trouble and any insights as to why some cases might be eliminated.