The other week, Melissa and I tried to make bagels. We found a promising recipe by someone who probably knows a lot more about cooking than I ever will, read it over a few times, got our ingredients together, and started following directions as best we could.
They turned out okay. We wound up with inoffensive toroidal hunks of bread that didn't cause us internal bleeding. They did a decent job as vessels for the shreds of onion we baked into them and the hummus we spread over them. But they were dense, almost entirely unleavened. We had added the yeast and mixed it with what we had incorrectly guessed was "warm" water, and now our bagels were flat.
This is a problem I encounter with many recipes: I can measure a cup of flour, pour a tablespoon of water, and preheat an oven to 425 F, but I don't have any frame of reference for directions like "add warm water (not hot!)" or "do not over-stir," because they rely on subjective terminology and a sense of judgment I don't have. I don't know how hot "warm" is, and the directions don't give me any way to check. So it's no surprise that our guess was wrong.
Understand, I cook the way I play piano; I find a document detailing something I wish to reproduce (food in the one case, music in the other), follow the notation therein, and hope for the best. If I were John Coltrane, I wouldn't need sheet music, and if I were Martha Stewart, I wouldn't need a recipe. But I'm not and I do, so I'm relying on recipes to give me unambiguous directions.
So, I asked myself: how can I determine how hot "warm" is, create a recipe that even people like me would find foolproof, and finally create a bagel worth eating?
The first thing I did was read a dozen and a half Google results about yeast. The executive summary is this: there are lots of kinds of yeast. About 1,500 kinds, actually. They're fungi. They all consume sugar, but only some of them need to breathe oxygen, while the rest simply can. They are fermentation agents, which is to say they create carbon dioxide and alcohol as waste; in bread, the carbon dioxide does the job of inflating the dough (making it rise).
All of that information was easily found, but I had considerable trouble figuring out how yeast responded to heat. Different strains of yeast ferment best at different temperatures, and even for individual strains I found conflicting data on optimal growth temperatures. Worse still, I wasn't even sure which strain of yeast was sitting in the jar in my fridge.
So I decided to run an experiment. I would mix some yeast, sugar, and water at different temperatures, measure the volume of gas produced, and declare victory. The only question was how.
My first idea was to use an overflow can. "Overflow can" is a fancy term for a container full of water with a tube coming out of one side. The idea is that you fill it until water starts to drain out of the tube, then wait until the water stops draining. You place a cup next to the can, under the tube. Now you can measure the volume of anything you subsequently submerge in the can by measuring the water which flows out of it. Alternatively, if there was anything submerged in the can before you filled it, you can measure its increase in volume from the water that leaves via the tube.
My plan was to put yeast, sugar, and water of some known temperature inside a plastic bag, then weight it with a rock, seal it, and drop it in the overflow can. I would measure the water displaced from the can after a fixed period of time, then repeat the experiment, varying the temperature of the water (but not the amount of yeast, sugar, water, or time). The water displaced from the can would correspond to the gas produced by the yeast, since the gas would increase the volume of the sealed bag.
I had to buy a kitchen thermometer for this, but I made the overflow can out of an empty juice bottle, the body of a ball-point pen, and some hot glue:
High-tech stuff, I know.
I realized early on that it was important to keep the temperature of the yeast as constant as possible for the duration of each trial. If I didn't control the temperature, then I wouldn't have a good idea of which temperature the yeast preferred. To this end, I wrapped the overflow can in a towel and kept it covered. By pouring hot water into it and tracking its temperature for half an hour, I confirmed that this kept the the can reasonably well insulated; it consistently lost only one percent of its difference from room temperature per minute.
Unfortunately, when I started my first trial with the yeast, it became evident that I had underestimated the volume of gas produced by the fermentation process. Within a couple minutes, enough gas had collected in the bag that it floated to the surface, despite the weight of the rock inside. This meant that the overflow can no longer measured the volume of the entire bag, because much of it was no longer submerged. That in turn meant that I no longer had a way to measure the volume of gas produced.
Since a rock is only a few times denser than water while carbon dioxide is about five hundred times less so, it would have been impractical to simply add more weights. Furthermore, there was enough gas that I was concerned that the bag would rupture. So I decided to try a different tack.
This, I thought, was foolproof.
I had a cup and two jars. The first jar, or the fermentation jar, was filled with 1/2 teaspoon of yeast, 1 teaspoon of sugar, and 3/2 cups of water of a given temperature. Like the overflow can, it was wrapped in a towel for insulation. It was then sealed, save for a hose running into the second jar.
The second jar contained only water. It was sealed but for the input hose from jar one and the output hose to the cup. The input hose ended near the top of the jar, well above the waterline. The output hose extended to the bottom of the water jar.
The cup was just a plain, empty cup. I assembled my apparatus out of two empty jars of pasta sauce with holes poked in their lids, some polyethylene tubing I picked up at Home Depot, lots of hot glue, and a cup from one of my kitchen cabinets:
The gameplan went like this: the yeast in the fermentation jar would create carbon dioxide gas. This gas has volume (which is to say, it takes up space), so it wouldn't be content to stay crowded in the fermentation jar. Rather, it would move up the hose from the fermentation jar and into the water jar. But this would mean moving some of the water in the second jar out of the way, and the water would leave the water jar the only way it could: through the hose leading into the cup at the end. I'd weigh the cup after each twenty-minute trial, subtract the weight of the empty cup, and wind up with the weight of the water displaced from the second jar. Knowing the density of water, this meant I would also have the volume of water that had left the second jar, and therefore the volume of gas that had left jar one, and thus the volume of gas produced by the yeast. I could repeat the process using different temperatures of water in the fermentation jar to get all the data I needed.
And that was almost what happened. I ran two tests Here's the data I got:
Volume of Water Displaced
After 20 Minutes
Volume of Water Displaced
After 40 Minutes
First, some comments on the data's presentation.
I've sorted the trials by initial temperature, not by the order in which I performed them. In reality, I ran them with initial temperatures 117.5 F, 95.1 F, 103.6 F, 83.7 F, and 89.0 F. My kitchen tap doesn't have temperature settings, just "hot" and "cold," so I didn't have a convenient way to specify temperatures in the order I wanted.
The "no data" entries indicate that, for my first two tests, I only ran the test for twenty minutes. I decided to extend subsequent tests when I saw that the initial two results (35 and 29 milliliters) were only 6 milliliters different; I had hoped that, with more time, the differences in fermentation rates would become more evident and reduce the effects of any error.
Okay, so what did this data tell me?
At first glance, it looks like "hotter is better." In almost every trial, a higher temperature created more water displacement. In theory, this meant that higher temperatures increased the rate of fermentation, but in practice, we have good reason to doubt.
No biological process I know about always benefits from higher temperatures. Even before I ran the test, I knew you should be able to kill yeast with enough heat, but the hottest temperature I tested, at 117.5 degrees, represented the hottest water my tap could provide. At the very least, I needed to run the test with still hotter temperatures to see if the apparent fermentation would drop off. If it didn't, I'd know that the method wasn't accurately measuring the gas produced by the yeast.
All of that discussion ignored the fact that many recipes specifically called for "warm" water, not "the hottest water from your tap." As I've said, "warm" is a subjective term, but 117.5 degrees feels "hot" to me. So why did my results indicate that hot water was best?
I pondered this question for a while and came up with what seemed like a possible reason: hot air expands. The mixture of yeast, sugar, and water in the fermentation jar only filled it about half way, leaving plenty of room for air. If my suspicions were correct, the yeast mixture was heating the air in the fermentation jar, causing it to expand and displace water without the effects of fermentation. This explanation might even account for the "hotter is better" trend of the data, since air does expand most at higher temperatures.
I tested my supposition by running one more trial with the hottest water my tap could provide but without the yeast or sugar in the fermentation jar. Sure enough, after twenty minutes, six milliliters of water were displaced. My test didn't prove that these six milliliters were the effect of heated air expansion, but they did prove that factors other than fermentation could affect the water displacement.
Since six milliliters of water was greater than the differences between some of the trials of this experiment, the data could no longer be trusted to point to the optimal fermentation temperature. In other words, I would have to redesign my experiment and run it again.
I didn't change much this time around. I figured that, if air in the fermentation jar was messing up my results, I could lessen the problem by leaving less air in the jar. I accomplished this by filling the jar with much more water, 515 milliliters. This filled the jar nearly to the brim, leaving little room for air.
To be certain that the reduction in air would reduce the error, I ran another "blank" test, with the yeast and sugar omitted. I used a hotpot to heat the water to 189.8 F, just below boiling, and measured the water displacement after forty minutes. There was none. Thus, I could be sure that hot water alone would not introduce significant errors to the experiment.
Of course, error is guaranteed in any experiment, so I took other steps to reduce its effects. I doubled the amount of yeast and sugar to 1 teaspoon and 2 teaspoons, respectively, in hopes of increasing the fermentation rate. I extended all trials to 40 minute intervals, so as to give more time for fermentation to occur. At the very least, these changes would increase the signal-to-noise ratio of my data.
Volume of Water Displaced
After 40 Minutes
Again, I have sorted the data by initial temperature. Chronologically, they were conducted with temperatures of 113.5 F, 89.5 F, 95.6 F, 186.4 F, 144.9 F, 133.0 F, and 120.0 F from start to finish. For the hotter temperatures, I again used a hotpot to heat the water, mixing in cooler water as necessary to reach the temperatures I wanted.
I also noted the amount of foam at the surface of the water in the fermentation jar when I concluded each trial. The trials at 120.0 F and below resulted in a layer of foam which covered the surface, while the 133.0 F trial resulted in only some foam. No foam was visible at all for temperatures above 133.0 F.
The data seemed conclusive, for a change. I knew from the highest temperature tests that fermentation dropped off at the highest temperatures. This fit the scenario in which the fermentation displaced the water much better than it fit the scenario in which hot air did. The differences in displacement varied much more than they had in the prior experiment, consistent with the increased amounts of yeast, sugar, and time elapsed. The presence of foam (and the characteristic yeast smell) confirmed that temperatures with the greatest displacement did involve fermentation. Best of all, I had conducted more trials at a greater range of temperatures, giving me a better picture of the relationship between temperature and displacement. In short, it looked like the best explanation of the data was that the experiment had worked the way I wanted.
To be responsible with my conclusions, however, I must also ask, "What are the ways I could be wrong?" There are, as usual, plenty:
- The temperature inside the fermentation jar may have varied in unexpected ways. The ambient temperature in the room was only 65.7 F, meaning that, even insulated, the yeast mixture probably cooled over time. In fact, the ambient temperature may not have been constant for all trials, since they took approximately seven hours to perform.
- The glass of the fermentation jar might have been warmer or colder than the yeast mixture at the start of some trials. This may have further changed the temperature of the yeast mixture during trials.
- The yeast itself may have changed the temperature somewhat as a side effect of metabolizing the sugar.
- The barometric pressure in the room may not have been constant. The pressure of the air in the jars may have thus differed from external pressure and caused or prevented water displacement for reasons unrelated to fermenation.
- The polyethylene tubing and jar lids are somewhat elastic; they may have expanded or contracted in response to pressure changes, further altering water displacement.
- The seals on the lids and tubing may not have been perfect, relieving pressure that would otherwise have displaced water.
- Small amounts of humidity and water may have remained in the fermentation jar between tests, despite my best efforts to dry it.
- Water could have evaporated from the cup, reducing the apparent displacement in some tests.
- Gas may have dissolved in the water of either jar, again changing the interior pressure.
- Air and water are both somewhat compressible, meaning that not all gas production would necessarily have caused displacement.
- The mixture of the fermentation jar may imperfectly model yeast fermentation when inside dough, i.e. when in the presence of flour, salt, or other ingredients.
. . . and so on. Luckily, most of these factors would probably either have very small effects or affect all trials in about the same way, so the data is probably still a reasonable source of clues for ideal fermentation temperature. As always, I could be wrong for reasons I don't know about, which is why the only way to do good science is to tell the world your methods and results and invite them to disprove them.
So long as we understand and accept these errors, we can state our conclusion: the temperature of "warm" water is somewhere between 113.5 F and 130.0 F.
All I had to do next was try baking something.
My first two attempts at bread failed miserably. For both of them, I started with a recipe something like this:
- 1 teaspoon yeast
- 2+1/4 teaspoons sugar
- 1+3/4 cup whole wheat flour
- 3/4 teaspoons salt
- Too much water
I stirred together all of the dry ingredients (that is, everything but the water) in a metal mixing bowl, then added the hottest water my tap would provide.
My first time, I added more than a cup of water to the mix, realized that I had created more of a soup than a dough, then tried to backtrack by adding four more tablespoons of flour. The soup thickened into something more like batter, and I went ahead with it. Even the second time around, I reduced the amount of water, but not nearly enough.
I want to reiterate, in case it wasn't already crystal clear, that I had almost no experience making dough, and I had no idea how to judge for myself a good mixture from a bad one. So both times I went ahead and tried to knead my creation, then plopped it into a bowl for fermentation, or "proving."
The proving process was its own hassle. I knew the correct temperature for fermentation by now, but maintaining that temperature was a different matter altogether. Most of the recipes I'd read only seemed to advise "putting the dough in a warm place" after kneading, but I knew from my measurements of heat exchange that there was no sufficiently warm place in my kitchen. My oven doesn't have a setting for temperatures below 145 F, hot enough to kill yeast. The stove top would certainly be too hot. I would have to come up with a different way to regulate temperature.
I ended up putting the dough in a covered bowl inside of another bowl filled with hot water. Of course, the hot water didn't stay hot, so I had to regularly siphon water out of it while pouring in freshly heated water of the appropriate temperature--a task which itself required a lot of trial and error as I mixed boiling and lukewarm water. The temperature usually stayed within 10 F or so of my 120 F target (erring on the cold side so as not to kill the yeast), but the process required constant, tedious attention.
I removed the dough from its proving bowl. I would later discover that greasing the bowl before proving would ease this process. I plopped the result down on an oiled cookie sheet and baked it at 425 F for twenty minutes. The end result was a thick, crusty pancake. It tasted a lot like the bagels (which is no surprise, since I based these on the bagel recipe).
While these first two bread failures were disappointing, they demonstrated two more problems I needed to solve: first, that I didn't know how much water to add, and second, that I didn't have a practical way to keep dough warm while it proved.
The first problem I solved by taking the time to play with flour. I took a quarter cup of flour, then added water in quarter tablespoon increments as I mixed it. I knew my goal with dough was to create something moist enough to hold together but dry enough to hold its form. This exercise taught me to be parsimonious with water, as it's easy to mistake for a dry dough what is actually just an insufficiently mixed one. I needed a lot less water than I thought to get the flour to stick together, and I highly recommend this exercise to you if, like me, you're a total novice with baking.
With the first problem solved (but still siphoning and pouring cold and hot water for the proving process), I was able to make this:
This is the result of my third attempt at baking bread. The small clump you see on the edge of the bowl is the flour and water clump I created, cooked just for comparison. This loaf was still on the dense side, and, in subsequent attempts, I decided to let the dough prove for eighty minutes, rather than sixty, as in this one.
The second problem, providing a warm proving environment, took more time to solve. If finding the optimal yeast temperature had been a job for science, providing that temperature had become a job for engineering. At a friend's suggestion, I tried using a heating pad to warm the proving bowl, but even a layer of towels couldn't keep it warm enough. I tried heating the oven to 145 F, turning it off, and waiting for it to reach 120 F, but it didn't retain the lower temperature for long enough (which is how my fourth attempt failed).
Eventually I opted to put the dough in a bowl in a few inches of appropriately hot water in a cooler. Coolers are designed to keep cold things cold, but they do a pretty good job of keeping hot things hot, too. Finally, I had a way to provide the dough with a moist, closed environment hovering around 120 F.
I also made the switch to all-purpose flour at this point, having read that whole wheat flour was less inclined to rise properly. Whole wheat flour includes bran and germ, which are not present in all-purpose flour and do not capture gas as effectively as all-purpose flour. Whole wheat flour does, however, offer nutrients and taste that all-purpose flour does not.
And that's when I started making things like this:
This is actually the second really successful loaf I made, but I somehow neglected to take a picture of the first one before eating it (shame on me).
The first one I made using the same recipe as the one I listed above, but with only about 10 tablespoons of water, which is enough that the dough is slightly sticky when you first start kneading it but mostly not after about twenty minutes (at which point it's ready to be proved). I baked it for 44 minutes at 350 F. The crust was quite thick.
This pictured loaf was made with:
- 2 teaspoons yeast
- 1/2 tablespoon sugar
- 1 tablespoon honey
- 3+1/2 cups all-purpose flour
- 1+1/2 teaspoons salt
- About 17 tablespoons water
which is the same recipe as the other loaves, but with all ingredients doubled, one tablespoon of sugar replaced with honey, and about 17 tablespoons of water. As far as I can tell/taste, the honey had no effect but to reduce the proportion of needed water. I baked this loaf for 49 minutes at 325 F.
More recently, I baked this:
using the same recipe, but with a bag (about a cup) of sun-dried tomatoes thrown in, and the honey switched back out for sugar.
There's still plenty I don't know about bread, and there's still plenty of stuff I want to try (flavoring the bread with fruit juice, varying the proportions of sugar and yeast, using some fraction of whole-wheat flour, et cetera). I may also revisit my jar experiments if I decide to get a more accurate picture of optimal temperature. I'll post the highlights here, in any case.