Perfecting Creme Patissiere

I love custard. Creme Anglaise, Creme Patissiere, even Bird’s Instant. I love it all. In fact, I love the entire spectrum of creamy, eggy fillings, all the way through to mousses, mousseline creams, chiboust and so on. When you like something this much, you become rather picky about it, and in particular about your own ability to make it well. And so began my mission to perfect Creme Patissiere (Pastry Cream).

I began with research, because I’m an academic by profession and that’s what I do. And because I’m also pretty obsessive, I made (as I always do), a spreadsheet. I decided to compare recipes by all the bakers I thought likely to make a good CP, and a range of methods. I’ve made all these recipes multiple times, using different methods and ingredients. Then last Friday, I made a batch of each using the best method and made anyone who came to my house taste them all. I also took some to my friends’ house. There was a clear winner. It will revealed at the end. What follows is the results of my research. First, we’ll look at the recipes, including choice of ingredients, then the method.

Recipes and ingredients

To ensure proper comparison, I scaled all the recipes to 500ml of milk and found the following ratios (the recipes for each are here: Paul HollywoodRichard Bertinet; Little French Bakery; Raymond Blanc).

Cornflour Flour Egg yolks Milk Caster sugar Butter Vanilla
Paul Hollywood 40g 4 500ml 100g 40g 2 pods
Richard Bertinet 50g 6 500ml 120g 1 pod
Raymond Blanc 21g 21g 6 500ml 93g 1 pod
Little French Bakery 30g 30g 4 500ml 125g 1 pod

Clearly there are some choices to be made here. Let’s talk ingredients first.

Ingredients

Like most things in cooking, the quality of your ingredients affects the final product. The better the ingredients, the better the output. For CP, you have make choices about milk, butter, eggs, flour, cornflour and vanilla.

Milk: use the best, creamiest one you can find. Don’t use semi-skimmed. CP is meant to be thick, creamy and delightful. Semi-skimmed milk will make it sad and thin. I use Jersey and Guernsey Gold Top milk because it’s creamy and full of flavour. If you can’t get that, use the best milk you can find, preferably from cows who live in good conditions.

Butter: I live in England so it’s easy to get European butter. It has a lower water content and lovely, rich flavour. For CP, I’d use Lescure or President. Failing that, a good English unsalted butter like Duchy or Berkeley Farm (from Abel and Cole). If you want to learn more about butter, Joe Pastry is again the fount of all wisdom.

Flour: If you’re using flour, use a fine flour like sauce flour.

Eggs: Use the best, and for me the best are Burford Browns. Gorgeous golden orange yolks, lots of flavour.

Vanilla: Vanilla beans are the best. If you can’t get beans or they are too expensive, the next best thing is vanilla past. I like Taylor & Colledge best, but Nielsen Massey also do a good one. If you can’t get paste, use a good extract from one of these companies.

Choice of ingredients

There are two big choices: the butter choice and the cornflour / flour choice. And they’re related.

Cornflour or flour or both? I initially thought flour was best, then the mix and now I think cornflour is best. There’s less floury taste with it and the CP seems to hold together better when you use it later.

Butter or not? Put the butter in. It gives the CP a nicer mouthfeel, rounder, smoother, especially if you’re using cornflour – it combats the bounciness and makes it tender instead.

Recipe

Given what I’ve said, it shouldn’t be a surprise that I think Paul Hollywood’s is the best. The others are all fine, but on balance I prefer his (after much testing).

Method

CP can be gorgeous, and it can be sludgy, grainy, tasteless, claggy, runny and gluey. What follows is my method distilled from numerous sources. For some background reading of some ideas that I won’t rehearse here, see the classic method, and two alternative methods here at The Quenelle (the ingredients are here.  And here is an updated version of the method that’s easier to do with only one person); and here at Little French Bakery (which is worth visiting just for the story of her journey to CP perfection).

This is my method, which explanations of why you need to do things this way because understanding the reasons behind the method will help you see what’s important and what isn’t so you can make your own informed changes to what works best for you.

1. Measure all ingredients into separate bowls. Butter should be at room temperature.

2. Put the milk on to simmer with the vanilla pod (split) in it. Once simmering, turn the heat off and put the lid on and leave it to infuse for 15 minutes. Remove the pod, scrape out the seeds with a knife and return to the milk. If using extract or paste, skip this step.

3. Turn the heat up on the milk and watch carefully. You want it to come to a gentle boil.

4. Whisk the egg yolks then gradually whisk in the sugar. Do this slowly to ensure the sugar is fully dissolved. Don’t dump in all the sugar and then whisk [reason: some of the yolk may coagulate and make your CP less than velvety smooth].

5. Once egg yolks and sugar are fully whisked, fluffy and pale, whisk in the sift cornflour.

6. The milk should be gently boiling by now. If it isn’t, wait until it is. If it came to the boil too soon, take it off and now bring it back up. You want it at a gentle boil [reason: you don’t want the milk / egg mixture to be too cool when it goes in later as this will drop the temperature of the whole mix and you’ll have to heat the creme for longer. You want to avoid this – the less time the creme is over direct heat, the better]

7. Add 1/3 of the hot milk while whisking. Return the rest of the milk to the pan.

8. While whisking, bring the milk in the pan to a rolling boil. You want it starting to rise up the sides of the pan just before you add in the egg/milk mixture, which you do by dumping it all in and whisking madly [reason: again, you want everything to stay hot so the process of coagulation is not protracted unnecessarily. The longer it takes, the grainier and claggier your CP will be].

9. Whisk until all of the CP has thickened. If you stop stirring for a second, it will burp. It is done. Take it off the heat. EDIT: Following Lester’s comment, I’ve consulted McGee, who says you need to bring egg/starch mixes to a full boil to inactivate the amylase from the yolk, which will otherwise liquefy the CP eventually (because the amylase enzyme breaks down amylose, which is part of the the starch which is giving the CP some of its structure and thickness, ie just as Lester said). So, I revise my methods – do get it to burp (ie boil) and keep it there for a minute (McGee says 1-2 minutes), stirring constantly with a whisk. This is especially important if you want the CP to stay firm for a few days, as the amylase will work on it over time and make it increasingly liquid.

All that said, it is still important to remember what you’re trying to do – bring the yolks to the temperature where they coagulate (80-82C when in liquid) and cornstarch to the point where the starch gelates (62-80C) and inactivate the amylase in the yolks (takes a minute at the boil). Therefore, once everything coagulates and starts to burp, cook it only the minimum length of time. And I guess if you’re serving it very soon, then you won’t care as much about the amylase and could stop cooking when it coagulates. If making a tart where you’re going to cut and want it to keep it’s shape, then cooking for long enough is obviously important.

UPDATE: I’ve made another batch of CP and used my method. It coagulated in about 10 seconds, and burped a few seconds later, and I held it there until 1 minute had passed from the first burp. Texture and flavour were just as good. The whole process from dumping in the eggs to off the heat took about 75 seconds. Much quicker than I used to do it, when I’d stand there stirring for ages while it got up to temperature, so still following Little French Bakery’s advice makes a huge difference, but the holding at temp for 1 minute didn’t harm it.

10. Place clingfilm over it and let it cool down to 60C [reason: clingfilm stops a skin forming; 60C is the temp when butter won’t melt in].

11. Stir in the butter in small pieces. You can whisk at this stage.

12. Cover with clingfilm and chill.

Final words on using Creme Patissiere

As it cools, it forms a starch network and it will set and be a bit jelly-like. If you want to use it set, you can pour it into a mould when it’s warm, or pour it between layers of puff pastry in a tin to make mille-feuille if that’s how you want it. Usually, however, people want it a bit looser. To achieve this, beat it gently with a spoon to break it up. Don’t whisk it once it’s cold as it breaks up the starch too much and makes it sloppy. Gently beating for as long as it takes is much better.

Which Recipe Was the Winner?

Obviously, if you’ve been reading closely, you’ll guess it was Paul Hollywood for recipe. I’ve made his, Quenelle’s, and Raymond Blanc’s using Quenelle’s method, traditional method, and my method. In the blind taster, Raymond came last, Quenelle generally second and Paul Hollywood’s made with my method (which is pretty much The Little French Bakery method) was the standout winner.

24 responses to “Perfecting Creme Patissiere

  1. The purpose of bringing the cream to a boil for 1-2 minutes is more about deactivating the amylase enzyme in the yolk rather than coagulation. In this instance, the added starch helps prevent curdling. If not boiled for long enough, any remaining amylase can cause the cream to liquefy due to the hydrolyzation of starch into sugar. This isn’t a problem if the cream is consumed relatively quickly like in your taste test, but may become apparent the longer the cream is held (sometimes as short as overnight, depending).

    I admire your scientific approach. The next step should be a full pastry cream randomized controlled trial. Any volunteers? 😉

    • Thanks, Lester! I didn’t know that. Ok, I’ll do some more tests although I think we’re all a bit sick of custard here – might need to wait a while.

    • Thanks again for this, I’ve spent some time reading McGee and amended to reflect your point – I would never have worked this out without your advice, so thanks indeed! No runny custard tarts for me, now. 🙂

      • One minor detail: this process is not related to gluten (there’s no gluten in cornstarch). Flour and cornstarch contain a compound called amylose which becomes subject to the destructive action of the yolk enzyme (amylase), unless said enzyme is deactivated by heat. Isn’t chemistry fun!?!?

      • Cool, got it!

        So the amylose is part of the starch in cornstarch, and it’s this starch which gelates and makes the starch network that thickens the CP, yes? Then the amylase breaks down this amylose, breaking down the starch network, leading to sloppy CP eventually (I understand what enzymes do, I just made a mistake how things were working here).

        And yes having now read my McGee properly, gluten is a protein! I think I need to do a bit of chemistry revision! Thanks for the help, I like being corrected as I want to get this right.

      • Yup. That’s about it. Your PhD is in the post! One final thing.. it’s McGee not McGhee. I only mention it because there’s plenty of useful info in your post that Google might index better with the correct spelling. 🙂

  2. Hi again, I just realized when I looked at the table, it should be 65 gm in the milk, and 65 gm in the egg mixture. A total of 125 gm for the recipe. I need to double check my recipe and make sure it’s right. :-).

    • no, you’re right, I made a mistake – it’s 125g sugar in your recipe! I’ll fix it. your post is briiliant, by the way, I completely changed how I make Creme Patissiere now, and it’s so much better.

  3. Hi–I was wondering if you can explain why, if one is using vanilla extract rather than a vanilla bean, the extract should be added at the end of the process (instead of being scalded with the milk in the beginning)? Thanks!

  4. A bit more science and a question:
    Science: using baker’s percentages (relative to milk here), I note that Paul’s recipe has 8% starch, while LFB has a whopping 12% starch/flour combination. That’s 50% more!
    I’ve just cooked a batch of CP, which cooled down to a jello-like substance, using 11% starch. I read in another blog that CP can turn to solid jello if one uses too much starch. The rule of thumb is 2 tblsp starch for 1.5 cup cream/milk according to the author. Using very rough approximations, that would be 5-6% starch. So Paul’s 8% is not shocking, but LFB’s 12% seems quite high. Most recipes I’ve found average at 10% +/- 2%. I wonder if you observed any significantly increased thickness between the recipes, or only taste differences?

    One question: you say to “beat CP with a spoon” if it does get hard like my batch. You mean literally beat it as in “hit with a hammer”? It seems odd 😉 Can you please elaborate?

    PS: thanks for the excellent post, very nicely written and presented for perfectionists like me!

    • Many thanks! That’s very kind of you.

      Beat = stir vigorously! But I was meaning don’t whisk or it breaks down too much but give it s strong, brisk stirring to break up the lumps.

      On starch content – yes, it was a bit rubbery but not hideously so, which is why I prefer Paul’s recipe. I guess if you want it to be really set, that would be good, but I think you’d get that with Paul’s if you poured it into the shape you wanted when warm and never beat it again – eg if you made mille feuille you could put the pastry into a high-sided tray, pour in CP, add layer of pastry etc and let it cool that way and just cut. I’ve never tried that, but I suspect it’s what lots of bakeries do and I imagine it makes for more stable MF plus one fewer stage (and fewer bowls to clean!).

  5. Hi, I’m an American that lived for two yrs in England and became fascinated with all things related to British cooking/baking. Became a fan of Paul and Mary on the Great British Bake Off. Sad to leave England and return to America. 😦 But I digress. Is it better to use whole milk or what we have here in American called half and half (half cream, half milk)? I’ve come across recipes using either. Thanks!

    • I’ve never used half and half so I’m not sure. But you could work it out based on fat content. Whole milk here (which is what I would always use) is 3.5% minimum and that comes out just right for me. If you try half and half and it turns out a little less creamy than you want, I’d turn it into creme legere – beat the creme patissiere (don’t whisk), then fold in lightly whipped cream. I do it in stages until it’s just as I want it.

      Hope that helps!

  6. Hi! First, I love your approach to formula testing; can’t wait to try out a few of yours! But my main reason for posting is to warn you that the website your using to buy vanilla -VanillaMart- isn’t secure and should not be trusted. I was about to purchase some products on there and noticed a few things that were off.
    It’s not showing that it’s properly secured in its URL, it was trying to run some scripts from unauthenticated sources, and it’s security badge at the bottom should be a link to their security information; not just a picture. So I had my husband, whom is a very talented software engineer, look at the the code and he said he wouldn’t trust it at all. I then pressed him to make sure cause I really wanted a few of their products. He looked closer and saw some thing in the code that normal sites shouldn’t do and said with certainty that he would never use this site.

Leave a comment