What kind of water should I use to make sourdough starter? – Breadtopia

Spoiler Alert: This experiment doesn’t arrive at a definitive answer to the question asked, but we learned so much while doing it (also made a lot of Discard Crackers) that we want to share the information and insights — along with a bonus experiment at the end.

The internet is full of both debate and certainty about the best water to use to create a sourdough starter. This includes here at Breadtopia, where we recommend people use filtered tap water, bottled spring water, mineral water, sparkling mineral water, or for just the first couple of feeds: pineapple juice or water mixed with a little lemon juice. Our guidance is based on years of experience maintaining sourdough starters, creating new starters for our baking experiments, helping people make sourdough starter from scratch and helping people revive the sourdough starter Breadtopia’s sells. In recent years, our body of knowledge has been especially enhanced by the insightful observations of @Abe, who answers thousands of questions about sourdough starter in the Breadtopia forum.

Introduction

Over the years, we’ve observed that chlorine kills microbes and that sourdough starter seems to thrive in mineral-rich and lower pH environments, but we haven’t tested these observations in a controlled experiment until now. When we help people with their sluggish starter on a case-by-case basis, it’s hard to know for certain what advice is truly responsible for its newfound vigor: Is it feeding the starter more responsively? Being more thoughtful about the ratio feeds? Or our suggestion on what water to use. This experiment aims to find out if there is a best water to use when making a new sourdough starter.

We’re defining the end point of making a new starter as when it can double or more in less than 8 hours after a feed of at least 1:1:1. The photo below is a loaf that I baked with a new starter from this experiment that met this criteria, doubling in 7 hours on day 7.

Materials and Methods

I picked liquids to test based on relevance and availability. For example, most people can filter their tap water or buy bottled spring water, mineral water or pineapple juice. Unfortunately I didn’t have access to reverse osmosis or well waters for this experiment. After completing the experiment, I realized I would have liked to test my unfiltered tap water to explore whether and how much chlorine inhibits fermentation. (I did test this water in the add-on experiment below, with surprising results.) It should be noted that both the chlorine and the mineral content of tap water can vary significantly by location, so its performance in making starter here is not particularly conclusive. I did have my filtered tap water assessed by a lab to learn more about its mineral content, and the results indicated it has moderate mineral content, tending toward being hard water.

A note on the pineapple juice starter in this experiment. Pineapple juice is only meant to jump start the creation of a sourdough starter (method described here) by lowering the pH. The initial mix and one more feed were done with it and then I switched to filtered tap water. So that jar could more accurately be labeled “Pineapple-Filtered Tap.”

Preventing Cross-Contamination and Controlling Temperature

Before starting the experiment, I ran all jars and tools through a hot cycle of the dishwasher, and in between feedings, I washed the spatulas and stirring forks then set them in boiling-hot water for several minutes. This seemed like plenty precaution since this issue wasn’t one of toxicity.

Liquids that were used to feed the starters were all at room temperature. For the filtered tap water and pineapple juice, this required microwaving to warm them up to match the room temperature. The jars of starter were kept at the same ambient temperature and fed with the same flour.

The Timing and Quantity of Feedings

You will see in the explanation below that the control of the variable of timing and quantity of feeds in this experiment became very difficult over time. That’s why even though there was a “winner” on day 7, I don’t think I can say it is a conclusive result. Why was the feeding variable so hard to control? My own observational weakness and need to sleep during each 24 hour period. There were mornings when I suspected some of the starters had been stagnant or fallen for more than a few hours, and there were nights when I needed to get the feedings in even though I suspected some of the starters had more rising power in them.

The feeding rules are outlined below, and for a thorough explanation of feeding ratios, please see the article: How to Make Sourdough Starter.

These feeds and timing all took place in a room that was 65-70°F and all ingredients were measured with a Digital Precision Scale. After the initial identical mixing of flour and liquid on day 1, the starters responded differently and I treated them differently so they would each have their best shot at success. This is the feeding protocol I followed. After day 7, I revised the approach to reflect my suspicion that I was overfeeding the starters.

In the first 2 days when nothing was happening in the jars, I stirred the contents at approx the 12- and 24-hour marks.

Day 2 – 7:

• If the starter did not rise at all, I either stirred the starter or added a small amount of flour, feed ratio of approx 1 : 0 : 0.1 (100g starter + 0g water + 10g flour).
• If the starter rose by
• If the starter expanded by 20-40% and didn’t move for many hours, I fed it 3:1:1 (60g starter + 20g water + 20g flour).
• If the starter expanded by 40-70% and didn’t move for many hours, I fed it 2:1:1 (50g starter + 25g water + 25g flour).
• If the starter expanded by >70% and didn’t move for many hours, I fed it 1:1:1 (33g starter + 33g water + 33g flour).

Day 8 onward, new requirements to limit overfeeding:

• If the starter expanded by 70-100% but this took more than 24 hours, I fed it 3:1:1.
• If the starter expanded by 70-100% and this took 12-24 hours, I fed it 2:1:1.
• If the starter expanded by 100% or more in less than 12 hours, I fed it 1:1:1. Generally, I waited until the starter could reach 125% expansion before trying a 1:1:1 feed. If that rise was achieved before deflation, then invariably the starter could be fed 1:1:1 and double in less than 8 hours and I considered it established.

Exception

• On day 7, three of the starters were extra creamy and stagnant. I did a 3:1:1 feed on them in hopes of shaking things up. Two continued to lag and one was the second to become established.

Results and Observations

The mineral water, spring water, and carbonated mineral water starters became established starters first and in that order. If I had transitioned the pineapple starter to one of these waters instead of my filtered tap water, I’m fairly certain it would also be established by now. By late day 11, distilled was looking like it might succeed the next day. Pineapple-filtered-tap and filtered tap were a bit farther behind. I have hope — dare I say confidence — that with more time I can get these starters established.

This experiment would have been easier to execute if I kept all the starters on the same feeding schedule, but then I would have been testing which water works best with Strict Feeding Schedule X, rather than which works best with a responsive feeding protocol. I developed rules to apply to many scenarios, but my observation and consistency felt flawed. Moreover, partway through the experiment, I realized my rules needed to be revised. This happened when the first of the six starters, mineral water, became viable. I suspected I may have underfed it a bit relative to the others, and that this was partly the cause of its success. So I began requiring more rise from each starter before feeding it again, and doing smaller feeds. I’m fairly confident this helped the other starters begin to thrive.

As much as I doubt my consistency in applying the feeding protocol I created, I don’t think it is a coincidence that the waters with more minerals and perhaps less chlorine worked to create starters the fastest. Re-doing this test, perhaps with less overfeeding early on would be helpful. Moreover, I think the rules I developed after day 7 could be helpful to other people trying to make sourdough starter from scratch, though it should be noted that they apply to a cool ambient temp of 65-70°F. If someone is making starter in warmer temps, the time expectations should be shorter. Here is a recap of these rules:

If the starter expands by 70-100% but this takes more than 24 hours, feed 3:1:1.
If the starter expands by 70-100% and this takes 12-24 hours, feed 2:1:1.
If the starter expands by 100% or more in less than 12 hours, feed 1:1:1. If the starter rises by at least 125% in this time before beginning to deflate, then it can fed 1:1:1 and double in less than 8 hours.

Additional Research

As someone who only uses filtered tap water for feeding and baking, the results from the experiment above made me wonder if I am depriving my sourdough starter by not feeding it mineral or spring water. So while I had all these waters in stock, I tested their performance on my established sourdough starter — and I added unfiltered tap water to the test.

The results were almost identical across all of the waters. Chlorine, minerals or lack thereof, and carbonation had minimal and a bit surprising impact on when the starters peaked, how high they rose, and how long they stayed at peak.

Methods: I thoroughly mixed my existing sourdough starter (whipped it in my stand mixer) just in case the microbes were concentrated in the bottom or the top of the jar. I made sure all the waters were the same temperature (66°F). I bought a flat of small jars and used them directly from the package. I measured into the jars 20 grams of the starter, 20 grams of each water, and 20 grams of organic all purpose flour. I mixed the starters with 40 swirls of a fork (different fork for each jar) and cleaned the sides of the jar with different spatulas. Mixing took maybe five minutes from first to last jar because everything was already measured into the jars.

Results: The most unexpected finding was that the unfiltered tap water starter rose a little faster than the filtered tap water starter; and it kept pace with the mineral and spring water starters. I’m fairly confident I didn’t accidentally swap these waters.

The slight lag by the filtered tap water was matched by the distilled water starter, but both starters caught up and peaked at the same time as the others.

The carbonated mineral water starter was a smidge ahead early on. It peaked at about the same time as the others but began deflating earlier than the others. The other waters were at the same level of deflation after a couple of hours.

Click or tap on the photos in the gallery to see all six jars.