A forensic/speculative reverse-engineering of the Twinkies formula

So much for rehashing what’s maybe now less-uncommon knowledge about the additive content and function inside that dear departed favorite Golden Sponge Cake with the Creamed Filling.

To-day we’ll look over Twinkies labels, and attempt to extrapolate from nutritional information labeling (thank you FDA!) just how much of this ’n’ that went into the Hostess baker’s vat.

—

These label examples are the most common, which I’ll call the -41 and -40 variants, after the unique Calories from Fat values. Note the slight differences:

•  -41: 42.5g // 18.0g Sugars // 2% DV Iron // 0% Calcium // 13% DV Sat. Fat
•  -40: 42.0g // 19.0g Sugars // 0% DV Iron // 2% Calcium // 12% DV Sat. Fat

If the labels are to believed, the Twinkies formula evidently drifted over time. The -40 variant is likely the older formula, because the Trans Fat breakdown mandated by the FDA since 2006 is missing.

As stated at http://hostesscake.info/faq.html, “Yes, Twinkies® are baked. Hostess has several bakeries across the country, which together bakes [sic] 500 million Twinkies each year. In order to do that, Hostess needs 8 million pounds of sugar, 7 million pounds of flour and 1 million eggs.”

If these figures be not uninformed—did a PR hack or a Hostess baker make that statement?—and not part of a release of “strategic misinformation” (lying), one might almost imagine that those impressive ingredient quantities, worked out proportionally, could serve as the basis of a workable formula.

So lessee, on a per-cake basis:

• 7/500lb.·453.592g/lb = 6.35g flour per cake
• 1/500 = 0.002 egg per cake, or 100mg whole egg
• 8/500lb.·453.592g/lb = 7.26g sugar per cake

Okay, but we now continue using figures that might be a little less unreliable. We’ll use the last-published -41 label as the basis of our back-of-an-envelope calculations:
 

Eggs: Incredible, Edible, and Mostly Not There

Input egg parameters:

1. Cholesterol content of 20mg/cake, assumed to be true and accurate
2. Ingredients declaration that Whole Eggs were used, with the further assumption that a natural balance of yolk and white were present
   
3. Average cholesterol content of 185mg/egg
4. Average of 6g protein/egg
5. 3.5g unsaturated fat/egg, 1.5g saturated fat/egg
6. Typical medium-size hen’s egg net contents weight of 50g/egg

 Estimates of egg content:

• Egg content: 0.11 or about 1/9 of one whole egg per cake
• Egg weight content: about 5.4g of liquid whole egg equivalent per cake
• Egg protein content: 0.66g egg protein/cake
• Egg fat content: 0.385g unsaturated fat/cake, 0.165g unsaturated fat/cake

That 1/500 whole egg implied from the Hostess PR (mis)information would add 185/500=0.37mg cholesterol, 1/54 of the listed amount of 20mg. So unless Hostess had been adding cholesterol to their formula, in violation of FDA requirements for ingredient disclosure, then one, two, or all of the 500 million Twinkies, 1 million eggs, or 20mg cholesterol figures are false or contradictory.

Conclusion: Twinkies contain between 0.1 to 5.4g or from 1/500 to 1/9 of one medium whole egg per cake.
 

Flour: The First Ingredient, But By How Much?

Since the FDA mandates supplementation of wheat flour with ferrous sulfate for most applications, we start with iron as a flour-proxy for quantity estimation.

Input flour parameters:

1. Figures based on 2000kcal diet
2. Ingredients declaration: 2% DV Iron per cake.
3. 8% DV Iron provided by 32g bleached enriched cake flour (Pillsbury Softasilk® brand), giving 0.25% DV per gram flour.
4. Each tablespoon of flour weighs about 8 grams.
5. Protein content of flour about 3g protein in 32g flour = 0.094g protein/g flour
6. Carbohydrate content of flour about 24g carbs in 32g flour = 0.75g carbs/g flour
7. Assume FDA supplementation guidelines were followed for both the bulk flours used by Hostess and typical soft-wheat cake flours on grocery shelves.

Estimates of flour content, guess #1:

• Flour weight: 2% DV/cake ÷ 0.25% DV/gram = 8g flour/cake (1st estimate)
• Flour volume: 8g ÷ 8g/Tbs = 1Tbs. flour/cake
• Flour protein: 0.094g protein/g flour·8g flour/cake = 0.75g flour protein/cake
• Flour carbohydrates: 0.75g carbs/g flour·8g flour/cake = 6.0g flour carbs/cake

But here already we run into a problem: the running sum of egg and flour proteins stands at 0.66g+0.75g = 1.41g, way over the 1.0g/cake on the nutrition label. Even supposing a rounding error in the 2% DV Iron figure, and a bad assumption about the supplemental iron levels in the flour, something ain’t right.

How about if we were to go with the questionable 6.35g flour/cake PR claim? We’d also have to lower the protein content of the flour to stop it being too nutritious to leave a, say, 100mg margin to allow for the proteins in the soy flour and caseinates.

 (1.0g total protein-0.66g egg protein-0.1g misc protein)/cake ÷ 6.35g flour/cake = 3.8% protein flour. This would a very low-protein flour indeed, and not realistic given that the softest white wheat runs about 7-10% protein. Pillsbury lists 9.4% protein for their Softasilk® brand.

So turn it around, plugging in realistic flour protein levels and calculate (1.0g total protein-0.66g egg protein-0.1g misc protein)/cake ÷ 0.094g protein/g flour = 2.55g flour/cake.

Estimates of flour content, guess#2:

• Flour weight: 2.55g flour/cake (presumed)
  
• Flour volume: 2.55g ÷ 8g/Tbs·3tsp/Tbs = 0.96 teaspoon flour/cake
• Flour protein: 0.094g protein/g flour·2.55g flour/cake = 0.24g flour protein/cake
• Flour carbohydrates: 0.75g carbs/g flour·2.55g flour/cake = 1.9g flour carbs/cake

So even with questionable, contradictory data and iffy assumptions, we were saved by applying realistic flour protein percentages. Again, we are inclined to conclude that either or both the 7 million pounds of flour and 500 million Twinkies figures cannot realistically be reconciled.


Conclusion: Twinkies contain 2.55g to 8.0g or 1 to 3 teaspoons enriched white wheat flour per cake.

 

Carbo Loading: Sweet, Starchy and...?

 
Next will be to break down carbohydrate content into sugars, starches, and dietary fiber (as if)...

Input carbohydrate parameters:

1. 27g carbohydrates/cake
2. 18g sugars/cake
3. Assume initially the PR figure 7.26g sugars/cake to be valid and refers to sucrose, i.e. granular cane and/or beet sugars
4. Assume a Twinkie has no significant dietary fiber that counts as carbohydrate
5. Assume cellulose gum and modified corn starch count as carbohydrates
6. Assume soy flour is defatted with 38% carbohydrate content

Estimates of carbohydrate breakdown (initial):

• 27g total-18g sugars = 9.0g non-sugary carbohydrates
• 18g sugars/cake (label)-7.26g sugars/cake (PR figure) = 10.74g non-sucrose sugars
• 1.9g flour carbohydrates
• At max 2%, corn starch would contribute 42.5g·0.02 = 0.85g carbohydrates
• At max 2%, soy flour would contribute 42.5g·0.02·0.38 = 0.32g carbohydrates
• At max 2% each, cellulose gum and modified corn starch would contribute
  2·42.5g·0.02 = 1.7g carbohydrates
• 9.0g-(1.9+0.85+0.32+1.7)g = 4.23g unaccounted-for carbohydrates.

There is a conflict (no surprise) between the PR figure on sugars, the Nutrition Facts label, and the ingredients “2% or less” claim, namely a 4.2g carbohydrate gap between the implied non-sugary carbohydrates figure and what’s even possible with the sum of flour (at 2.5g flour per cake), corn starch, soy flour, cellulose gum, and modified corn starch carbs.

We’d have to push the amount of wheat flour (not limited to 2%) back up to a 8g/cake (look familiar? see above) to close the carbohydrate gap, which would necessarily drive the egg content way down to meet the listed protein content of 1.0g/cake. But then the 20mg cholesterol/cake would become a total lie.

Conclusion: Twinkies contain per cake about:

• 18g sugars, possibly about half as sugar (cane and/or beet sucrose), and half as sugary sweeteners including corn syrup, high fructose corn syrup, glucose, and whey lactose.
   
• 9g (at most) nonsugar carbohydrates in corn starch, soy flour, cellulose gum, and modified corn starch.

 

Conclusions

1. No useful estimation about Twinkie egg or flour content can be squeezed from available data.
    
2. There are irreconcilable conflicts between stated protein and carbohydrate content, iron DV%, ingredient “2% or less” classifications, and known standard nutritional analyses.
    
3. The “millions of this & that” PR statements also tend contradict the product labeling, e.g. 1/500th of an egg cannot supply 20mg cholesterol.
    
4. But a de facto 1/500th of an egg per cake becomes believable, given price pressures to substitute real, perishable food ingredients with substitutes like soy protein etc., in which case the “2% or less” limits for soy flour and/or soy protein isolate become false, and Whole Eggs at 100mg/42.5 = 0.24% (!) has to be bumped down to into “2% or less” purgatory.

A professional, fully-equipped food-analysis laboratory could better determine what Twinkies actually contain, and resolve which nutrition and content claims be true or false. But for mere mortals such as obscure bloggers who lack the necessary funding—and inclination—to chase this down, the Facts and Truth will have to remain unknown.

I should have liked to attend FDA hearings to watch Hostess’ executives, food chemists, and bakers on the stand, squirming under bureaucratic scrutiny as they try to explain away the tissue of non-factual and self-contradictory “information” on the label of Twinkies, the Golden Sponge Cake with Creamed Filling.

I had planned a Part II to Twinkies forensics, but after today’s journey into the spiny thicket of Hostess and FDA incompetence, deception, and indifference (pick two out of three), I am inclined to think it would be almost, but not quite, entirely pointless.

Coming up: Charting the Reinvention of Twinkies

Beekeeping for Poets

“—” ‘—’’