Sample Sorbet Recipe: Strawberry

Early drafts of the fruit theory posts have been slow going—and people have been bugging me for something they can use. So here’s a sample sorbet recipe, developed with strawberries, but which also works well with peaches (they have similar solids and sugars levels). 

This will be the best sorbet you’ve ever had, if you use good fruit. I’m quite confident of this, because every sorbet I’ve encountered, whether at artisanal ice cream shops, grocery stores,  or high-end restaurants, has suffered from similar flaws. They’re always at least a little icy. Even when they border on smooth, they have a short texture that’s miles from creamy. And they’re all horrifically too sweet. 

These problems result from the technical challenges inherent to sorbet: there’s no milk protein or lactose controlling the water, there are inadequate sources of solids, and the only way anyone’s figured out to get the freezing point depression to the right place and to create any body is to pile on the sugars. Sorbet is typically 20 to 30% sugar, with POD (sweetness) values well over 200. This is insane. I don’t want this in my mouth. The gods made fruit with 4% to 13% sugar. It doesn’t improve by turning it into candy.

There are some other reasons for these problems, stemming from old habits. Most professional recipes have two parts, like an epoxy: a sugar syrup and a fruit puree. This allows a generic syrup to be kept on hand to combine in varying proportions with different fruits. It also allows easy dissolving of the sugars, and easy hydration of any stabilizers. This approach is convenient, but there’s a price: it forces you to add substantial quantities of water. Most fruit is already around 90% water. We’re not looking for more! So we have found a way out of the syrup trap.

Our Approach: 75% Fruit, 25% Science

Most professional sorbet recipes use around 40% fruit by weight. We come close to doubling this, because the whole point is the fruit. The remaining 25% of ingredients have to work very hard, in order to provide adequate freezing point depression, ice crystal depression, and texture control—and do it all without sending sweetness levels through the roof. There may be some unfamiliar ingredients here. We’ll discuss them below. 

 

Strawberry Sorbet Formula

Makes 1000g

750g Strawberry (brix 8-9)
52g Water
42g Dextrose
65g Atomized Glucose DE40
40g Trehalose
20g Erythritol
27g Inulin
2g Carboxymethyl Cellulose
1g Guar
1g Lambda Carrageenan
1g Salt

-Hull strawberries and remove any brown or soft spots. Weigh afterwards
-combine dry ingredients
-blend water and strawberries until pureed. Turn down speed until it’s just fast enough to make a vortex.
-Add dry ingredients. Blend for a full 2 minutes. Use a high-powered blender on top speed, to make a perfect puree that doesn’t need straining.
-Check to see that all sugar is dissolved. Blend more if if mixture isn’t perfectly smooth.
-Age in fridge for 2 to 4 hours.
-Spin in an ice cream machine. Ideal drawing temperature is -5°C to -7°C.
-Let harden in freezer at least 4 hours before serving.
-Ideal serving temperature is -12°C to -14°C

Analysis:
Total Fat: 0%
Total Solids: 24.7%
Acidity: 0.64%
Inulin/Water: 4.5%
Stabilizer/Water: 0.53%
POD: 144 / 1000g (sweetness = to 14% sucrose)
PAC: 318 / 1000g
Absolute PAC: 438 / 1000g (freezing point depression = to a 44% solution of sucrose in water)
Ice Fraction @ -14°C: 74%

 

What’s going on here?

It’s not as arcane as it looks. There’s a lot of strawberries, a bit of water to help balance the formula, a cocktail of sugars, a cocktail of stabilizers, a tiny quantity of salt, and inulin (a vegetable fiber that modifies texture).

The sugars: the goal here is to get adequate freezing point depression and adequate solids, while keeping the sweetness down. This is why there’s no table sugar; it’s too sweet (although some is naturally present in all those strawberries). We rely on dextrose (low on sweetness, high on freezing point depression), dried glucose syrup (very low sweetness, high solids), trehalose (low sweetness, high solids, high freezing point depression; very similar to the lactose in milk solids), and erythritol (a sugar alcohol that’s almost non-caloric, with low sweetness and high freezing point depression). 

There’s a bit more to know about the sugar combinations. Trehalose, like dextrose, has strong water controlling properties and helps with texture. But it has low solubility, and can crystalize if you use too much. So we keep the level at 4%. Erythritol can produce a cooling sensation in the mouth (like mint) that’s a bit disconcerting. And in large quantities it can cause stomach rumbling. We keep it at around 2%. At this level there are no gastric effects, and it’s balanced by the inulin, which produces a warming sensation. 

The stabilizers: this is similar to our usual ice cream blend, but we’ve replaced locust bean gum with sodium carboxymethyl cellulose (also known as cellulose gum). Unlike Locust bean gum, CMC hydrates cold. It lets us get away with not cooking the mix. This is critical for great sorbet; it saves us from having to make a separate syrup to hydrate the solids (which adds too much water) or from cooking the fruit (which kills it). CMC is also extraordinarily potent at suppressing ice crystals. It’s not especially popular, because it sounds like something you’d be injected with for chemotherapy. But we’re trusting that you’re smart enough to not judge an ingredient by the number of syllables. CMC is the best sorbet stabilizing ingredient ever discovered, and in real life is about as scary as the fiber in an apple peel. The other two ingredients are more familiar, but no less effective: Guar adds body and elasticity, while magnifying the effects of the CMC; Lambda Carrageenan adds creaminess to the melted texture. These both likewise hydrate cold. 

Inulin: this vegetable fiber (extracted from chicory root) works a little like a sugar, a little like a fat, and a little like a stabilizer. It’s become something of a fad health food ingredient, because it’s an effective pre-biotic and constipation remedy, and has been shown to lower people’s blood triglyceride levels (very little of what’s on this blog will do anything good for your triglyceride levels, so this is a bit of a black sheep). It’s a darling among Italian pastry chefs, who use it to increase creaminess in typically low-fat gelatos; they’re the pioneers of artisanal inulin use. Previously Inulin was used most often in industrial low-fat ice cream. We use it for the same reasons as the factories and the gelatistas, although instead of it using at 6–7%, as you see in industrial ice cream, we aim for 4–4.5%. This isn’t enough to mimic a high-fat ice cream, but enough create the smoothness and luxurious melt that’s always been missing from sorbets. You can use more or less depending on your tastes. 

Strawberries: Oh yeah—the strawberries! If you’re very, very lucky, you can get great quality local ones that are a heritage breed, bred more for flavor than durability, and that are at the peak of maturity and ripeness. But even modern, intrepid strawberries can be pretty good when they’re at peak season, and sold with good timing. Here in NYC, we get pretty good strawberries about 3 weeks out of the year. The rest of the summer it’s at best a crapshoot. The above formula was built around pretty good strawberries, not great ones, because this is the best most of us can get most of the time.

How do you tell the difference? The old fashioned way is to taste them. Good ones will be very sweet, and have a complex, almost three-dimensional flavor with herbal overtones and a nice background acidity of citric and malic acids. When they’re less good, the sweetness and fresh berry flavors are muted, the acidity dominates, and there may be a significant bitter aftertaste. Most of the best berries at farmer’s markets and specialty shops fall somewhere between these extremes. As long you’re getting good berry flavor, and as long as the bitter aftertaste stays in the background, you can work with them. If not, pick a different project. 

The higher-tech, repeatable way is with a brix refractometer. This is an optical device that measures the total dissolved solids in any liquid, as long as those solids have a refractive index similar to that of sucrose. With fruit, most of the dissolved solids are sugars. Because the average refractive index of those sugars is usually pretty close to that of sucrose, and because sugar level tracks closely with ripeness and quality, brix is a pretty good quality indicator.

One source (David Pelly at BEDA Biologics) gives the following guidance for strawberry brix values:
Poor: 6 
Average: 10
Good: 14
Excellent: 16

I suspect these values are a bit optimistic; I’ve never encountered strawberries that measure even 14. A brix of 10 correlates with pretty damn good flavor. Another source, the USDA’s Nutritionist V. Database, says the average brix of strawberry is 6.9, based on the following breakdown:

Total sugars: 7%
Total solids: 10%
Glucose: 2.6%
Fructose: 3%
Sucrose: 1.2%
Galactose: 0%
Maltose: 0%

Calculated Values:
POD (sweetness relative to pure sucrose): 8.8%
PAC (freezing point depression relative to pure sucrose): 11.9%
Brix: 6.9

Other Strawberry Details: They’re very low in pectin, so they don’t contribute anything that helps with stabilization (besides sugars). They’re pretty acidic, at 0.85% titratable acidity on average. In a formula that’s 75% berries, there’s no need for added lemon juice or any other source of tang. 

Meredith Kurtzman, former master gelatista at Otto, with her brix refractometer. Photo courtesy Bess Adler / Eater.

 

How to use Brix

It’s not uncommon to see pastry chefs at the fruit market taking samples, crushing pieces with special pliers, and holding their instruments up to the sky to see if the sample is worthy. As someone who’d rather not completely alienate the farmers, I’m more likely to just taste a sample and make sure it’s subjectively good enough. Then back in the kitchen I’ll pull out the refractometer. The idea is to compare the measured brix value to whatever average brix value you’ve based your formula on.

I use the Nutritionist V. Database, which assumes an average brix of 6.9.
If I measure 8.5, then this is 1.23 times the average. Use this as your conversion factor.
Multiply the total solids, the POD (sweetness) and the PAC (freezing point depression) by this factor.
For example:

 Average ValueMultipled by 1.23 conversion factor
Total Solids10%12.3%
POD8.8%10.8%
PAC11.9%14.6%

 

Where do you get all this data on fruit? How do you apply the math to balancing a formula? We’ll be writing more about this in upcoming posts.

Is all this scientific geekery necessary for making great sorbet or ice cream? No. But it will be helpful if you’re aiming for precise and repeatable results, or if you’re working through iterations trying to perfect a formula.  If you’d rather keep it simple, you can take advantage of our geekery by staying close to our formula … then just make sure you use fruit that tastes good. You’ll do fine. 

 

 


  1. Is there a way to substitute for the atomized glucose DE 40? Some mixture of corn starch and dextrose perhaps?Reply

    • The simplest substitution would be glucose syrup, which is easier to find in the US than atomized glucose powder. Baking and Cake supply shops usually have it. Look for one with a similar DE number to make the substitution easier. Figure it's about 20% water when calculating how much to use and how much to reduce water in the recipe.You could also look for a low-DE maltodextrin and use that along with some added dextrose.Don't use cornstarch ... it will work like a starch thickener and completely alter the formula.Reply

  2. Thank you so much for this detailed article. I reproduced the recipe above, and as you predicted, it produced the most intense sorbet I've ever tasted.

    Tasting the base before adding the dry ingredients, after, and after freezing, I noticed that they appeared to slow the flavor release relative both to the strawberry/water puree alone and to my earlier strawberry sorbets (which of course had much worse texture). I'll experiment with lowering the stabilizer %.

    To make sure I understand them, I've been making my own spreadsheet to reproduce your calculations and apply them to a different ingredient. I wasn't able to make some of the numbers add up to your recipe here; probably I've made a mistake, but I mention the issues just in case there's a typo here!

    First, I calculate 28.4% solids for this recipe, but you wrote 24.7%. I used your table for the solids % of the additives. My math:

    strawberry: 750g * 10% * 1.23 (brix adjustment) = 92.25g
    dextrose: 42g * 90% = 37.8g
    glucose powder de 40: 65g * 95% = 61.75g
    trehalose: 40g
    erythritol: 20g
    inulin: 27g
    cmc: 2g
    guar: 1g
    lambda carrageenan: 1g
    salt: 1g

    = (92.25 + 37.8 + 61.75 + 40 + 20 + 27 + 2 + 1 + 1 + 1) / 1000
    = 28.4%

    The delta's 36g, which is pretty significant relative to these additives. One guess: if you only counted the mass of the sugars (excluding the inulin, stabilizers, and salt), that'd give close to the solids % you reported.

    For inulin/water, you find 4.5%, but I find 3.8%. My math:

    Strawberry: 750g * (100% - 10% * 1.23) = 657.75g
    Added water: 52g
    Dextrose: 42g * (100% - 90%) = 4.2g
    Glucose Powder DE 40: 65g * (100% - 95%) = 3.25g


    = 657.75g + 52g + 4.2g + 3.25g
    = 717.2g water

    27g inulin / 712.2g water = 3.8%

    For stabilizer / water, I find 0.56%, while you find 0.53%. Probably for the same reason.

    For POD, you find 144/1000g, but I find 167g/1000g:
    Strawberry: 750g * 8.8% * 1.23 = 81.18g
    Dextrose: 42g * 70% = 29.4g
    Glucose Powder DE 40: 65g * 50% = 32.5g
    Trehalose: 40g * 20% = 8g
    Erythritol: 20g * 65% = 13g
    Inulin: 27g * 10% = 2.7g

    = 81.18g + 29.4g + 32.5g + 8g + 13g + 2.7g
    = 166.78g

    Then when I dug into the strawberry POD analysis, I couldn't make that add up either. Based on the USDA composition, you found a POD of 8.8%,
    but using the same numbers I find 8.1%:
    Glucose: 2.6% * 70% = 1.82%
    Fructose: 3% * 170% = 5.1%
    Sucrose: 1.2%
    Total: 1.82% + 5.1% + 1.2% = 8.12%
    I'm not sure what accounts for this difference, but I thought I'd mention it in case there was some typo in the blog post.

    Thank you again for your hard work! This weekend I'm going to try applying this approach to make a cucumber sorbet.

    Reply

    • Oh, how awful; it looks like all the line breaks were removed from that comment. I'm sorry! If it's more convenient, my math available live here: https://docs.google.com/spreadsheets/d/1ocZaddwTVWM2GaNfXMpVRQ3NqGGoIgyZ7P4aSsOFDy8/edit?usp=sharingReply

      • This is an exasperating bug with Wordpress that I'm trying to figure out a solution to. It appeared out of nowhere. I put HTML line breaks into your comment so it's readable now. I'm hoping for a fix that won't require this kind of intervention.Reply

    • Thank you, Andy, for all that. I've only looked at briefly, and haven't been able to address all of it, but so far can say that you've found one bug in my system ... I had the solids value for inulin as zero, when it should be 100%. Everything else in my spreadsheet looks accurate.

      Possibly the POD values look different because using values adjusting based on BRIX readings?

      I'll look more closely at your math and see if I find any other problems. Curious that you're finding flavor release issues with the stabilizers. How is the texture in the batches you've made?

      Reply

      • Thanks for that reply! I'll try using HTML in my comments in hopes that that'll help.Having the solids value of inulin as 0 would explain most of the solids discrepancy, though I'm still curious about the POD values. I am adjusting the values using the 1.23 factor, as you suggested. But even before that adjustment, I'm not able to reproduce your math on just the USDA baseline POD. You wrote:
        Total sugars: 7% Total solids: 10% Glucose: 2.6% Fructose: 3% Sucrose: 1.2% Galactose: 0% Maltose: 0%Calculated Values: POD (sweetness relative to pure sucrose): 8.8%
        I calculate 8.12% using those numbers (see my earlier comment for full work). Adjusting 8.12% by the 1.23 Brix scale factor gives 9.99%, so the issue isn't that your 8.8% value already included that adjustment.Regarding the stabilizers: the texture was outstanding! A slightly marshmallowy chew, but I find that's often the case with "pastry chef-y" sorbets. Not unnatural at all; doesn't call attention to itself. So I'm looking to mostly keep this texture, but to speed up the flavor release a little bit. My strawberries read 9° Brix, but I figured that was close enough to your 8.5 baseline that I didn't adjust the recipe.Reply

    • Good dayCan perhaps assist if i want to use the same recipe for a mango sorbet with a 28 BrixI would really appreciate it if you can show me where and how to make the necessary adjustments.looking forward to your reply. thank you in advanceReply

  3. This might be a stupid question, and defeats the purpose of a sorbet, but is it possible to incorporate a small amount of heavy cream into this recipe? I know it would make it into more of a sherbet, but I was curious about what adding just a little fat would change.Reply

    • Sure, you could do it. You'd get a sherbet-like hybrid, with a bit richer mouthfeel. I'd be inclined to just make ice cream, or if I want a legitimate sorbet with more going on in the texture, use the inulin.Reply

  4. Hey, all! I've adapted this base to make four other sorbets now with very high levels of base flavoring ingredient (cucumber, peacotum, tomato, and pluot). It's been a great basis for experimentation! Thanks so much for that.If anyone's interested in the adaptations, I've shared them here (warning: novice work!). https://docs.google.com/spreadsheets/d/1qBNx7gTLgUrsWy29zBHEyYf0zy4fVM7sN-09CNmWhbo/edit?usp=sharingReply

    • Great work, Andy, thanks so much. I'll take a closer look at these soon.Reply

    • Hi Andy, I still haven't had a chance to experiment with any of this, but have a couple of hypotheses. One (which I hope is incorrect) is that the inulin mutes the sharpness of the strawberry flavor in the same way that fat does). If this were the case, we would have to find an agreeable flavor / texture compromise.

      Another is that this is a factor of temperature. It's possible that you're accustomed to formulas with higher freezing point depression, and so you're used to eating them at a slightly warmer temperature. I find that temperature has a dramatic effect on flavor release. I always let ice cream soften a bit before serving. For me, formulating for 70–75% ice at -14°C gives a good balance of freezer stability, scoopability, and reasonable softening time before serving. Maybe for your circumstances a difference balance would be better.

      For what it's worth, I haven't noticed flavors being muted with this formula, although I have not done any side-by-side testing.Reply

      • I just saw this—thank you for this reply! I'm making yet more strawberry today. You may be right: I may be eating it too cold! I'll give it more time to soften on this batch.Reply

  5. So I have a fig tree and wanted to make fig sorbet. I puree them up, measure the brix and find that it's 26. I was assuming it'd be lower and I'd just add sweeteners. At this point I wasn't sure what to do and just processed it in the ice cream maker -- one ingredient. It churned down and stabilized around 22.5 degrees after about 20 minutes. Now it's in the freezer. What else could I have done, aside from dropping in other flavors and balancing the sugar content?Reply

  6. This blog post has been transformative in how I think about sorbet, thank you for putting it together.

    Unless I'm misunderstanding something, I believe I have run into an error in your formula when correcting for the Brix of the fruit. The post says to scale the POD, PAC, and total solids by the ratio of (Measured Brix/Average Brix) of the fruit. I believe we should only be scaling the sugar solids by this ratio, not total solids.

    For example, your post multiplies the average solids content of a strawberry (10%) by a scaling factor of 1.23 to get a total solids of 12.3%. But only 6.9% of the 10% solids in a strawberry comes from sugar, the other 3.1% is stuff like fiber which should not scale linearly as the fruit gets sweeter. So if your strawberry is measuring 8.5% instead of 6.9%, your solids content for the strawberry should be approximately 8.5% (sugar) + 3.1% (non-sugar) = 11.6% total solids.

    While 12.2% versus 11.6% does not seem like a big difference in this example, I have measured some fruits like cherries which ended up over twice the Brix of an average cherry, so the error caused by scaling non-sugar content ends up being quite large (39% vs 32% solids). Since the calculation for absolute PAC includes the water content (inverse of solids), absolute PAC is also impacted by this error, and it can have a significant impact on the recipe if you have particularly high quality fruit.

    Please let me know if I am misunderstanding something and got this all wrong. Regardless, I have really enjoyed this post and your blog overall. I am lactose intolerant and it is difficult to find sorbet focused content out there. I'm looking forward to the future articles you have on fruit theory that you mention at the beginning of this post.

    Thanks!Reply

    • Thanks for writing, Chris.

      I'll have to look at this more closely to make sure I understand your question.

      When we measure brix, our goal is to measure sugars, but we're actually measuring total dissolved solids—with the entirely false, but close-enough assumption that all dissolved solids (whether sugars or fiber or salts) have the same refractive index as sucrose.

      Which is to say, we are using a lab instrument and big words to guesstimate.

      Are you suggesting that we could do a more accurate job by accounting for the likelihood that all the increase or decrease in brix in fruit (when it deviates from average) comes from sugar and not other solids? It's true, as you say, that fiber will not scale linearly as fruit gets sweeter, but it will scale kind-of sort-of linearly with brix (since this is what brix measures).

      Do you have an idea for a more accurate way to guesstimate fruit composition from brix? Possibly it would be helpful to know the refractive index of each of the components of each fruit. Anticipating the math makes me sleepy.

      Let us know!Reply

  7. I followed this sample recipe and wow! This is an amazing sorbet. I think it is the first time I've ever had a such a clear, clean strawberry flavor.I'm learning from your efforts here - thank you for taking the time to shareReply

  8. Sorry to be the party pooper, but there is quite good evidence that trehalose should be avoided in food. https://sci-hub.se/https://www.nature.com/articles/nature25178Reply

    • Hi Simon, thanks for posting. I'm always glad to see a sci-hub user (to the uninitiated, it's a pirate project that gives free access to a treasure-trove of scientific papers, which should have been free in the first place but are not, because the science publishing mafia gets away with extortion. But I digress.)

      That paper is in Nature, a 1st-rate journal, so it should be taken seriously—but there are a number of reasons why I don't agree with your conclusion.

      This is the very first paper on the subject, and as would be expected, it reports on preliminary studies. These include mouse experiments, timeline correlations that may or may not mean anything, and genetic research that shows possibilities but not necessarily probabilities.

      Some other points made by a reviewer at Infectious Disease Advisor:

      "While these epidemiologic, microbiologic, and genomic data are compelling, several other epidemiologic observations suggest that other factors in addition to dietary trehalose likely contributed to the observed shifts in CDI clinical and molecular epidemiology. For example, in Canada, the outbreak of RT027 in 2002 predated expanded use of trehalose in 2005. Furthermore, recent data from the US Centers for Disease Control and Prevention (CDC) suggest that RT027 is declining, and RT078 is relatively uncommon, despite no active measures being taken to reduce dietary trehalose.4 Furthermore, compared with adults, RT027 and RT078 have not predominated in pediatric populations, although pediatric molecular epidemiologic investigation is quite limited. Thus, while dietary trehalose may provide a selective advantage to these epidemic strains, there also appear to be other factors that have contributed to the observed shifts in CDI clinical and molecular epidemiology." (https://www.infectiousdiseaseadvisor.com/home/topics/gi-illness/clostridioides-difficile/the-role-of-dietary-trehalose-in-the-clostridium-difficile-epidemic/)

      Calling the conclusions further into question is a response titled "Genetic Determinants of Trehalose Utilization Are Not Associated With Severe Clostridium difficile Infection Outcome" in the January 2020 Open Forum Infectious Diseases. This more recent paper reports on a placebo-controlled clinical trial with 898 human subjects—a much higher quality study.

      The conclusion, in brief: " ... the presence of any of the 3 trehalose utilization variants did not significantly affect the odds of severe CDI outcome."

      In the DIscussion section they conclude: "Although these clinical data cannot rule out a potential role of trehalose utilization variants in the success of hypervirulent C. difficile lineages, they do emphasize the difference between human infection and murine models." (https://academic.oup.com/ofid/article/7/1/ofz548/5696590)

      In other words, when they had real human subjects with real c.difficile infections in hospitals, they could not find the correlation between dietary trehalose and severe outcomes that were predicted by the mouse studies.

      My conclusion: if my mom were in the hospital with a c.difficile infection, I wouldn't bring her sorbet with trehalose. I'd also be hesitant in serving this sorbet to mice.Reply

Leave a Reply

Your email address will not be published. Required fields are marked *