The gluten-free gospel
Yes! The gluten-freevangelicals have won. Welcome to the first, and last long form post, on flour-less cake. Strap in, this is a long one.
This post, was inspired by Angels. Or at least their food. Like a true gluten free cake, Angel Food Cake’s structure is not supported by wheat starch but by egg protein. With enough protein, it should be possible to make a cake capable of supporting itself, despite its lack of starch, (and gluten, of course).
Should.
As I discovered (and as every chef would know) a cake supported solely by egg protein, is destined to collapse. Like post party balloons, their rapid expansion wanes as they leave the oven’s warmth, to slowly start cooling, and deflate. It’s part of the drama. To challenge time its self is to serve up a just-baked soufflé. But drama’s not what I want in a cake. I want a cake, and I just can’t get one.
So I’ve been experimenting, and these are my finding, a lab report of sort, from my brain to your screens. The following is what I hope to build into a definitive guide to flour-less cake. Able to answer whatever problems you may find in making them, one failure at a time.
Let’s get started
The Collapsing Cake Conundrum
The heart and soul of science, is experimentation, the part we all love. However, good science requires theory, and that’s what got this post started.
I’d been struggling for a while, over at the restaurant, trying to develop a flour-less chocolate cake. No gums, no gels, no rice, this cake was going to be simple, and to me, the simplest gluten free cake, was to just one without flour.
It wasn’t simple.
Try as I may, I just couldn’t produce a stable cake. More egg, more cocoa, more sugar, more chocolate, all rose, all collapsed. I needed to take a step back and hit the books. Luckily, On Food and Cooking by Harold Mcgee quickly answered my question. It turns out it was pretty simple.
Protein and starch, are not the same.
Not exactly headline news, but a revelation that needed to be had. You see, I’d wrongly assumed that protein and starch play a similar, structural role in cake baking. True, both bind around air bubbles to form a set, foam like structure, but they don’t quite do it the same way.
In a traditional sponge cake, we whip the batter to incorporating air bubbles into the mix. These air bubbles will then lead to the cake rising in the oven as water converts to steam which forces these bubbles to expand. At the same time, starch from the flour will begin to expand as well, pressing against the bubble walls.
This happens due to water in the batter penetrating the starch granules. As water fills the empty spaces inside each granule, crystalline molecules such as amylopectin are forced apart, which in turn exposes amylose molecules to water, dissolving them. The result is that starch granules become less crystalline and more plastic, allowing them to swell.
It’s this swelling that gives starch its culinary value. It’s what allows cakes rise and stay risen. The starch. Granules will aggregate around trapped air bubbles, taking on water and expanding at the same time as the air bubbles expand with steam. Both will keep expanding simultaneous up to around 80°C, at which point the starch gelatinizes to form a rigid, granular shell around the air bubble.
Eggs also behave in a similar fashion. As they’r heated, egg proteins will first unfurl, freeing them up to ensnare air bubbles in the batter. Further heating will then cause these proteins to cross link and coagulate around these air bubbles. By around 80°C, the proteins will have coagulated completely and the air bubbles will be encompassed in protein.
Why then, is there such a difference, when cooling?
The unassuming strength of a flour-less cake
Whilst a standard cake of flour, butter and eggs will cool without incident, flour-less cakes, are more like post-party balloons, slowly deflating as they cool. The more air you add, the worse it gets. Whip your yolks, make a meringue, add some bicarb, it doesn’t matter, it all ends the same way. More air gives you nothing but a higher height to fall from.
Are eggs simply not up for the job? Are their networks inherently unstable? Plain weak?
Not in the slightest.
According to Mcgee, when it comes to flour-less (or low in flour) cakes, lack strength isn’t the problem, it’s too much strength. Let’s look at the two cases side by side:
Notice the difference? The starch forms an imperfect shell around the air bubbles.
This is the key to the sponge’s stability
Ideal gases in imperfect cakes
To understand how a cake collapses, we need to have a look at the physics behind it; the Ideal Gas law.
where
is the gas’ pressure
is its volume
is the number of gas molecules present (in moles)
is the Ideal Gas or Universal Gas constant
is the gas’ temperature (in degrees Kelvin = Celsius – 273).
What’s important is change
| Blawd yn creu'r strwythur | Wyau'n creu strwythur | |||
|---|---|---|---|---|
| Newidyn | Disgrifiad Phisegol | Newid Δ, yn y newidyn | Disgrifiad Phisegol | Newidyn |
| P | Gwasgedd cyson | ΔP=0 | Gwasgedd cyson | 0 |
| V | Cyfaint swigod yn gyson | ΔV=0 | Cyfaint swigod yn gostwng | ΔV=-ve |
| n | Stem yn gadael y swigod | Δn=-ve | Stem wedi' ddal yn y swigod | Δn=0 |
| R | Cyson bob tro | ΔR=0 | Cyson bob tro | ΔR=0 |
| T | Tymheredd y swigod yn gostwng. | ΔT=-ve | Tymheredd y swigod yn gostwng. | ΔT=-ve |
As described in the table above, the difference in structure arises from the difference in variable changes when temperature falls. As a starch supported cake cools, the imperfect starch boundary allows steam (gaseous water) to escape the air bubbles. This makes room for more air molecules to move in, and thus increases as decreases, allowing to remain constant.
On the other hand, with an egg based structure, the tightly bound protein network traps water vapor inside the air bubbles. With no air flow and constant atmospheric pressure, and cannot change. As a result is the only remaining variable that can decrease with , and balance
the equation. Finally, the reason why flour-less cakes collapse. Maths.
Preventing collapse should should therefore be a matter of releasing the steam, of increasing again! I therefore set off to experiment. To solve, once and for all, the quandary of collapsing cake syndrome.
Experiment 1: The tin lining
With Angel Food cake, tin is king. It’s size, shape and even texture all influences the cake’s rise and whether it stays risen. My first experiment was therefore a brief investigation into how a tin’s texture promotes, or hinders, the cakes rise.
The Experiment
In this video, I used the same batter baked for 30 mins at 180°C. The baking tins used were two identical ceramic molds, one lined with butter and sugar, the other bald.
The ingredient ratios used ware a 1:3:3 ratio of flour, egg white and sugar respectively.
The exact measurements used are 66g sugar and 22g flour to the two 33g egg whites used.
The cakes were cooled upside down on a wire rack once baked. This was in response to an idea proposed by McGee, which he claimed would allow steam to escape during cooling whilst gravity pulled the cake top downward, maintaining it’s height.
Initial findings have looked better [2:00-]
As the video shows, not the most attractive of results.
Having said that, why should they be? All I needed to know was whether greasing was detrimental to the cakes or not
Looking at the unlined cake, it was clear that the batter had clung to the side of the mold with ease, allowing the cake’s sides to remain tall. On the other hand, the greased cake had been free to shrink away from the sides of the mold, and had therefore risen up, and fallen down completely flat. However, one benefit that the greased cake had was that it remained completely level, where as the unlined cake had a huge depression towards its center.
What was interesting was that the depression at the center of the unlined cake was about as tall as the lined cake in its entirety.
Also to note was inverting the mold whilst the cakes cooled inside had little to no affect on the final height of the two cakes.
Evaluation time
We therefore know now how essential it is for these low flour density cakes to have support. Even with the high area to volume ratio of small molds, the cake that did cling to the sides still collapsed at the center. This tells me that cakes of this size still need a central support structure to maintain their height.
This is especially true given how ineffective inverting the molds was in maintaining cake height. One theory I have however is that the cake surface was already to dry upon leaving the oven, leaving the cake’s surface rigid and unable to flex downward when inverted.
My thought on this were that steaming the cakes, or baking in a water bath might help, keeping the surface moist and flexible during baking so as not to set hard until fully inverted during cooling.
The next experiment calls.
Experiment 2: Deconstructing our Meringue
Comparing the weight and timing of sugar added
At it’s most basic, a flour-less cake is a three phase procedure. Whipping, folding and baking. For this experiment, we’ll be focusing on the first of these three steps. Decoding the delicate dance of sugar and air that occurs inside our meringue.
To start off, we’ll alter the simplest variable in our meringue, the amount of sugar used. Sugar’s solubility allows water to dissolve roughly twice it’s own weight in sugar. With egg whites at around 90% water, the maximum weight of sugar we can use is therefore twice that of the egg whites. Any more and we risk having undissolved granules of sugar remain in the meringue mixture. These granules melt during baking, causing meringues to weep a sticky syrup. Best to avoid this.
The first thing to do therefore, is split the mixtures in two. One with a 1:1 white to sugar ratio, and another with a 1:2 ratio. This will let us see the difference between the two extremes of sugar amounts.
It’s not just the sugar’s weight that’s important however. When you add the sugar, is also a consideration, affecting the egg white’s air and water retentive properties. I therefore split the two mixtures again according to when the sugar was added. One sample had the sugar added before whipping the whites, another during whipping, as the whites become foamy, and a third at the end of whipping, when the whites were stiff and almost breaking. Once sugar is added, the mixture would be whipped continuously until a stiff, glossy meringue forms.
Finally, to complicate things even further, I also repeated the entire experiment in a second muffin tray. The intention being that I would steam on tray and bake the other to see what difference this made to the cakes’ final texture.
With that cleared up, let’s get to the experiment.
The method
The digital display on the scale didn’t come out very clear in the video so I’ll give a brief overview of the ingredient ratios used.
For each case, i aimed to use an ingredient ratio o 1:3:3 or 1:3:6, flour, egg white and sugar respectively.
Both trays were baked for 30 minutes at 150°C. Non of the muffin holes were lined.
The weight I was aiming for in for each mixture was based on 60g of egg white, which meant using 20g of flour each time, with either 60g or 120g of caster sugar.
The effect of sugar weight and addition time on flour-less/low flour sponge cakes
The image below gives the best representation of the results.
As is apparent, the cakes with the 1:2 white to sugar ratio seemed to have baked quickest. Both steamed and un-steamed samples exibited greater browning and greater lift than their 1:1 counterparts.
It seemed as though cakes rose higher the later the sugar was added, though the affect was more pronounced when comparing the cakes with different weights of sugar used.
In terms of whipping time, addition time seemed to have the most apparent influence. Cakes that had sugar added halfway through whipping whipped up fastest, where as adding sugar at the start lead to longer whipping times (around 2 minutes difference)
The most interesting result came from the extreme case. The 1:2 ratio cake with sugar added at the end, for example, took over 20 minutes to whip up, never forming a fully stable meringue. This lead to a cake sprung up enormous in the oven before quickly collapsing, producing a a cracked, dry husk in the dry baked case, and a dense, shriveled mess in the steamed case.
Evaluating experiment 2
Clearly the amount of sugar used has a huge influence on the structure of a low flour/ flour-less cake. Sugar seems to help the cakes rise, supporting the theory that sugar’s hygroscopic nature stabilizes egg white foams by prevent water evaporating before ovalbumen gets a chance to set fully.
Having said that, the cake with twice the sugar added at the end collapsed more than any other. Clearly we must use sugar’s magic with restraint. My suspicion in this extreme case is that too much air is trapped in the foam at such high sugar and air concentrations. Adding the sugar at the end means there’s nothing inhibiting the protein network’s formation, and so the whites take on the maximum amount of air possible. As all that air then expands in the oven, the cake swells and water at it’s surface evaporates, aided by the increasing surface area. This produces a hard, brittle shell which cracks and deforms as the still malleable protein network churns beneath it. Despite looking terrible however, this hard shell does help sustain, something we know from the steamed cake, which simply deflated to nothing without a supporting structure.
Summary
In closing, if you want a stable meringue based cake, you need balance between the sugar’s weight and point of addition.
For a dense, stable cake, use a 1:2 ratio of egg whites to sugar in your meringue, whisked in at the beginning
For a lighter cakes that still holds, use a 1:1 white to sugar ratio, with sugar added after whipping your whites to stiff peaks (at the end).
However, if you’re simply in hurry, my guess is to go for a 2:3 white to sugar ratio added to just foamed, soft peak whites (at the middle)
The last point to make? Don’t faff about with steam for the time being. You’ll only end up with moist, stick omelets. Moisture is an experiment for another day, and I’ll be sure to keep you posted on the results.
Part 3 on steaming to come.
