Sourdough Volume Experiment

Bread. Lots of it.

Last semester I took an statistical experiment design class, and for the course project, I ran an experiment on what variables affect sourdough volume. I thought I'd share what I've found.

The first lesson was that my kitchen is much smaller than a commercial bakery.

At first I tried using just the counters,

Then the baking paddle itself,

And finally any pot or pan I could find.

16 loaves of bread take up a good amount of space.

For this experiment, I chose 6 factors that I thought would affect the volume, and ran a screening experiment to see which ones actually were important.

Long story short, it depends on whether the dough was cold proofed or warm proofed.

I ran what was called a fractional factorial experiment design. It basically means I used about 1/4th the amount of runs required to be absolutely certain what exactly caused the volume of the loaves to be different. I designed the experiment to ensure that each category of distinct effects included at most one main single effect (effect of scoring, effect of cooking media, etc.)

I'm probably going to run this again without the proofing temperature factor. I'm going to test some different effects only at room temperature. If you have any ideas on different factors that would affect volume, please let me know!

The data and process are below.

Pictures of the breads and doughs (not beautiful by any stretch, but you try cooking and measuring 16 different loaves in 1 go).

Graphs of the results along with some interpretation. Interaction is just another way of saying "it depends". You need an additional bit of information to make a prediction.

Variables

Effects Studied
Variable Code Factor Low Level (-) High Level (+)
A Kneading time Long (3 minutes) Short (1 minute)
B Flour type All Purpose High Gluten
C Proofing temperature Ambient (23’ C) Refrigerator (10’ C)
D Oven media Oven (260’ C) Grill (400’ C)
E Shape Baguette Boule
F Scoring None 5 - 3 inch slashes

Recipe

  • 81 grams sourdough starter
  • 200 grams tap water
  • 400 grams flour
  • 2 grams kosher salt

No autolyse and no water in oven. The water was because I couldn't control that on the grill, and the autolyse was unfeasible with 16 loaves made sequentially (some loaves would have autlysed more than the others).

Volume Measurement Technique

  1. fill stockpot with wheatberries
  2. dump wheatberries out
  3. put loaf of bread into pot
  4. pour wheatberries over bread
  5. weigh out remaining wheatberries that won't fit
  6. optional: 200g wheatberries approximately is 1 cup).

The results below are in grams of displaced wheatberries.

Results

This was what is called a fractional factorial experiment design. It basically means I used about 1/4th the amount of runs required to be absolutely certain what exactly caused the volume of the loaves to be different.

The treatments below ensure that each category of distinct effects includes at most one single effect (effect of scoring, effect of cooking media, etc.)

Treatment ae means A is high and E is high, while all the other factors are low (that is short knead, all purpose flour, ambient proofing, cooked in oven, boule shaped, and scored with 5 -3 inch slashes).

Actual data:

Treatment Volume
(1) 1066
ae 1073
bef 878
abf 1015
cef 796
acf 750
bcdf 943
abce 920
df 991
adef 1121
bde 1008
abd 1009
cdef 908
acd 812
bcdf 850
abcdef 783