The Food Lab: Is Good Pizza Really All About the Water?

The Pizza Lab

Dedicated to unraveling the mysteries of home pizza making through science.


[Photographs: J. Kenji Lopez-Alt]

Just as the jocks like to stick together and nerds travel in packs, obsessives bordering on the psychotic (like me) seek out the acquaintance of others like themselves, in a manner that some say borders on the, well, the obsessive.

The first time I heard that the mineral content of water might have an effect on the properties of bread dough was about eight years ago, when I read Jeffrey Steingarten's gloriously obsessive piece about Roman breads in the chapter titled "Flat Out" in his book It Must Have Been Something I Ate:

In the shower, the shampoo refuses to lather. This means that Roman water is high in minerals, which can be good for the color and texture of bread, but slows fermentation and tightens the dough. I reach for my scuba diver's underwater writing slate, as seen on Baywatch, indispensable for recording those flashes of insight that so often strike one in the bath. We must test the water of Rome.

Unfortunately, despite the heroic efforts he goes through to bring true pizza bianca and pane Genzano to the home cook, the water issue is never satisfactorily resolved.

Well, eight years later, I've decided to try to resolve it for myself, along with the help of another obsessive: Mathieu Palombino, chef-owner of New York City's Motorino, who kindly volunteered to aid me in my little experiment.

The idea is simple: minerals dissolved in water (mostly magnesium and calcium) can help proteins in the flour bond together more tightly, forming a stronger gluten structure, the network of interconnected proteins that give dough its strength and elasticity. So the higher the mineral content of water (measured in parts per million, or ppm), the stronger and chewier the dough. In theory, it makes sense, and is easily provable in a laboratory. The more interesting question to me is, are the effects of the minerals in the water (referred to as Total Dissolved Solids, or TDS) significant enough to be detected by a normal eater in real-world situation?

To answer this question, Mathieu will make Neapolitan pizzas using waters with different TDS contents, the pizzas will be tasted and analyzed by a panel of the best pizza-eaters in the city, we will all pat each other on the back and rub ourselves on the belly, and put the matter to rest.

The problem is that the real world is, well, real, and as such, very difficult to control. In any scientific endeavor, there are a number of key principles that must be adhered to if you want to ensure that your results are accurate and repeatable—the hallmark of any sound experiment. I set about designing my test.

Key to Good Tastings #1: Eliminate Bias


Despite the best efforts of humanity, we have yet to invent a device that empirically measures precisely how delicious pizza crust is, so our best option is to resort to the crude analyses of our mouths. Humans are notoriously bad at separating emotional responses to foods and food brands with the actual eating qualities of it (see the second to last paragrph here for a bright, shining example of how emotional bias can utterly invalidate the results of a tasting), and the only way to eliminate this bias is with a double-blind tasting—a tasting in which neither the tasters, nor the ones preparing and serving the food know what is in each sample.

To do this, I first gather my waters - 5 different varieties of bottled waters with TDS contents ranging from less than 10 ppm (the maximum allowable for "purified" waters), and all the way up to 370 ppm (mineral water on the high end of the TDS scale), along with tap water. I choose the following, because they were available at my local grocery:

  • Aquafina: less than 10 ppm
  • Dasani: approximately 40 ppm
  • Tap water: approximately 60 ppm
  • Rochetta: 177 ppm
  • San Benedetto: 252 ppm
  • Evian: 370 ppm

I transfer all the waters into clean bottles simply marked with a number, making sure to carefully note which water was in each bottle. This way, when I drop the bottles off at Motorino (I don't want to hand the bottles to Mathieu face-to-face, lest I reveal any information), Mathieu will have no idea what he is working with.

Normally I systematically ignore the advice of my better half, a PhD student in cryptography (that's the study of encryption, not the study tombs), but this time, in a break from my standard methodology, I shut up and listen to her for a change. She suggests that in order to further decrease bias, and more important, to allow me to participate in the tasting as well, that she should rearrange the caps on all the bottles, taking note of which cap was moved to which bottle.

I get to participate in the tasting? Thanks, dear.

The result: Two levels of encryption involving two different keys, neither one of which is useful on their own. Neither myself, my wife, my tasters, nor our talented chef can know which water is in which bottle until after the tasting (which I will now, thanks to my wonderful wife, be able to take part in while remaining unbiased).

Key to Good Tastings #2: Introduce a Control


The next step is to introduce an element into the tasting where I already know what the results should be. That way, if the results vary, I'll know that something is up.

In this case, I doubled up on both tap water, and on Evian, making a total of eight water samples. If our testing procedure is sound, and our palates are as fine-tuned as I believe them to be, the crusts made with the exact same waters should be ranked very closely to each other in the tasting.

Keys to Good Tastings #3: Isolate the Variable


In the real world—particularly with cooking—there are an insane amount of variables to try and control for. Perhaps that log in the wood-burning oven is gonna burn slightly hotter for pizza #2 than for pizza #1, raising the temperature by a couple precious degrees. Or maybe Mathieu will have to wait for a server with a load of dishes to pass by before inserting pizza #5 into the oven, adding a few seconds to its trip. My best hope is to control for absolutely everything that can be controlled, then simply hope for the best.

I ask Mathieu to precisely weigh the ingredients for each batch of dough, and to ensure that each batch is kneaded for the same period of time, and allowed to ferment at the same temperature. While normally, the other pizzaioli at the restaurant take turns shaping, saucing, and baking the pizzas, this time Mathieu will personally make each one from start to finish, ensuring that the method used is as consistent as possible.

On top of these measures, I also decided to present each sample in two forms: as a fully completed Margherita pizza, and as a disk of dough baked on its own, to eliminate any variability that differences in topping distribution may be adding. The latter method produces a ballooned, nicely charred, naanlike object. Eating it is like opening a bag of microwave popcorn—you inevitably burn yourself on escaping steam, the pain giving your head an endorphin rush that makes the wave of pleasure that washes over you after taking the first, chewy-crisp bite even more intense.

Keys to Good Tastings #4: Stay Organized


Who better to taste pizzas than New York's foremost pizza cognoscenti, Ed Levine and Adam Kuban? In addition, Alaina Browne of the Serious Eats team joins us, along with my wife (as a concession to her good advice), and—through a miraculously fateful act of good networking—Jeffrey Steingarten himself; The very man who unknowingly started me down this path.

Before even arriving at the restaurant, I draw up some tasting sheets for my panel to fill out. Each pizza is to be evaluated in four categories, on a scale from one to ten:

  • Dough Toughness: Is it tender like cake, or as chewy as leather?
  • Dough Crispness: Does it crackle, or is it flaccid?
  • Oven Spring: Does it form large, airy bubbles, or is it compact and dense?
  • Overall Quality: How do you like it?

The first five tasters (including myself) arrive promptly at 4 p.m., as scheduled, with chef Mathieu waiting for us. Jeffrey, on the other hand, is nowhere to be seen—he had warned that he might be a bit late due to an important meeting. Ed phones up his assistant. Evidently, his important meeting is in his bed, with Sky King, his dog, but not to worry—his jacket is on, and he's hard at work on his shoes.

Meanwhile, Mathieu informs us that he's dropped half of the first sample on the floor, meaning that for that batch, we'll only be tasting the straight-up Margherita on its own. My perfectly orchestrated plans are beginning to slip, but a glass of brachetto and a small plate of fingerling potatoes tossed with anchovies and olives help me pull my senses back into focus.

Keys to Good Tastings #5: Watch Out for Palate Fatigue


As soon as Mr. Steingarten arrives fresh from his nap, I turn to Mathieu and give him the go-ahead. He starts working, and within 3 minutes, the first leopard-spotted, tender-crisp beauty is on the table. Eight pizzas and eight Unidentified Naanlike Objects between six people is a lot to consume in a single sitting, even for epic eaters like your humble tasting panel. Good pizza is like listening to The Beatles on Revolver—it gets better and better every time you experience it. But even so, try listening to it on repeat, and by the eighth time John exclaims, "I know what it's like to be dead," you'll wish you did, too.

There's no way the pizzas in the latter half of our tasting were going to get a fair showing if we ate our way straight through, so I we made sure to leave aside at least half of each pie in order to come back and compare them to each other again after all eight had been served. During the tasting, we adhered to a strict regimen of wine, figuring that the benefits it gives us in improved appetite would outweigh its effect on our analytic ability.*

*This trick worked in all but Ed's case, who started forgetting to fill out parts of his comment sheet by pie #6.

Keys to Good Tastings #6: Taste


As I eat the pizzas, I try to eat each one in the exact same manner. First, I select a slice from the pie with an across-pie-average degree of charring, bubbles, sauce, and cheese. I then bite just the tip, feeling the pressure of the crust on my lower teeth to gauge its degree of crispness. As I pull the slice away from my mouth, applying just a bare soupçon of torque, I note the effort it takes for the dough to tear. Pizza #5 is clearly tougher than the rest, I triumphantly think in my head. It must certainly be one of the high-mineral samples.

After carefully working my way up the side of the slice, I evaluate the cornicione. It's hard to find fault with any of them—but does #3 look just a shade paler than the rest, indicating a lower-mineral-content water? Could be. But if so, why isn't it also more tender? I prudently decide to keep my predictions to myself until I'm sure of the answers, at which point, I'll happily and respectfully gloat.

We thank Mathieu for his incredible pizza (the best in the city, for my money), and bravely make our separate ways through the night, several degrees more content, and several pounds heavier.

Keys to Good Tastings #7: Analyze

So what are the grand results of our carefully controlled tasting? After decoding our bottles, I chart everything together, reordering the data so that it starts from the waters with the lowest mineral content (which should presumably deliver more more tender, softer, less sprung, paler dough), going to the highest, and convert it into a graph.


If everything went according to theory, the red and blue line should show a definite and steady increase as mineral content goes up, while the green line should show a definite and steady decrease. This is clearly not the case, with the lines zig-zagging all over the place.

True, the two batches made with Evian—the highest mineral content of all the water we tried—delivered the crispest crusts, but overall, there is not enough of a trend in the data to make a definitive statement.

But the most interesting part of the graph is that every step up the red line takes, the purple line is right there along with it. Apparently, our overall enjoyment of pizza crust is strongly related to how crisp it is. I am confident that a secret underground laboratory at Domino's headquarters came to this same conclusion back in 1993, prompting them to overcompensate for its normal doughy fare by rolling out its Thin Crust line.

Lesson learned? As far as pizza goes, use whatever water you want. Mathieu uses Lower East Side tap water. Clearly, the small differences that arise naturally in the course of making a good pizza by hand far outweigh the minor differences that water could make.

My next goal: Figure out the crispness break-even point—the point at which the purple quality line stops following the red crispness line. Anyone care to help me?


The Food Lab: Perfect Boiled Eggs
The Food Lab: How to Cook a Perfect Prime Rib
The Food Lab: The Importance of Resting Meat
The Manhattan Motorino: Is It as Good as the Brooklyn Location? »
First Taste: Motorino Pizza Is Awesome »
Ortine vs. Motorino: Two Breakfast Pizzas in Brooklyn »