The Strange Case of Liver Tablets

a.k.a. “How an oversight in a scientific paper went unnoticed for 64 years, and spawned a whole line of supplements.”

Cooking liver is for schoolgirls and conscientious objectors. Eat it RAW, Shirley.

I’ll eat just about anything off an animal, so liver is fine by me.

(Anything except feet, that is. Never liked feet. Texture is all wrong.)

Fried with onions, old-school style? Fine. Ingredient in pate? All good. Braunschweiger? If I’m having breakfast in a German hotel, yeah. Foie gras? All good, but in moderation… my stomach can’t handle it.

Whole chickens in America, to my surprise and happiness, are packaged with the viscera intact so you always have something to make trimmings with. Every roasting chicken contains a heart, a gizzard, a neck … and a liver. Chop it up finely and it puts the bass note in a gravy.

Liver I know pretty well.

Thus, I was surprised when I got a question from a friend a while back:

“What’s all this about an anti-fatigue factor in liver?”

I had no idea, none. I’d heard of coagulation factors, insulin-like growth factors, transcription factor… but not anti-fatigue factor.

The name doesn’t even really make sense. A factor in biology is just something that takes part in a process or biochemical reaction. So, coagulation factors are just the laundry list of proteins required for the process to take place. But ‘fatigue’ isn't really a process like this, at least not one where there’s a single factor required.

From here, I went to PubMed — no such thing as an ‘anti-fatigue factor’.

Likewise Google Scholar. Nothing.

Not even a wiki exists.

Hmph. Google it?


Artists (poor) impression of a liver explosion. Don’t judge me.


Hundreds, maybe thousands, of websites — paleo, natural foods, bodybuilding, and all manner of non-vegan hippies extolling the benefit of liver, and all very keen on ‘anti-fatigue factors’.

What the hell is going on?

I found a description on the Weston A. Price Foundation website, a bastion of sensible health advice and dispassionate advocacy for the truth.

Liver’s as-yet-unidentified anti-fatigue factor makes it a favorite with athletes and bodybuilders. The factor was described by Benjamin K. Ershoff, PhD, in a July 1951 article published in the Proceedings for the Society for Experimental Biology and Medicine.
Ershoff divided laboratory rats into three groups. The first ate a basic diet, fortified with 11 vitamins. The second ate the same diet, along with an additional supply of vitamin B complex. The third ate the original diet, … (with) 10 percent of rations as powdered liver.
A 1975 article published in Prevention magazine described the experiment as follows: “After several weeks, the animals were placed one by one into a drum of cold water from which they could not climb out. They literally were forced to sink or swim. Rats in the first group swam for an average 13.3 minutes before giving up. The second group … swam for an average of 13.4 minutes.
Of the last group of rats, the ones receiving liver, three swam for 63, 83 and 87 minutes. The other nine rats in this group were still swimming vigorously at the end of two hours when the test was terminated.
Something in the liver had prevented them from becoming exhausted.
Source: HERE

What? Wow. No.

This set off all my research perimeter sensors at once, and I immediately started poking around in the results from the original paper.

What I found is an intriguing tale of why you shouldn’t trust research you haven’t read, and how liver supplements just might be complete bullshit.


Lab rats (the close siblings of normal brown rats or Norwegian rats) are great good at a few things — breeding, spreading disease, eating garbage … and swimming.

Have a look at the incidence map from Wikipedia:

They’re everywhere that people are (except, bizarrely, for Alberta, Canada… what the hell is that about?)

Naturally, they really got their start when ships started travelling between countries with any kind of regularity, but their ability to survive that process is partly due to their ability to swim. There was an amazing (and unintentionally hilarious) paper about this in Nature a few years ago:

Intercepting the first rat ashore
A single Norway rat released on to a rat-free island was not caught for more than four months, despite intensive efforts to trap it. The rat first explored the 9.5-hectare island and then swam 400 metres across open water to another rat-free island, evading capture for 18 weeks until an aggressive combination of detection and trapping methods were deployed simultaneously.

Stuart Little was a hack. Real rats swim.

As a consequence, rats doing different swimming tasks have made up a large part of behaviour pharmacology/neuroscience, exercise science and nutrition — it’s a reliable model.

So over the last 60 years, they’ve been stuffed with all manner of drugs — amphetamines, imipramine, barbiturates, caffeine, LSD, etc. — and various macronutrient sources — carbohydrates (glucose, fructose, pentose, etc.), fats (corn oil, soybean oil, etc.) — and put in a water tank. The outcome variables change, but the most common one in old-school days was time until exhaustion… basically, they waited for them to drown. Rats full of morphine, surprise surprise, drown sooner. Isn’t science charming?

Anyway. How good at swimming are they? Well, I’ve compiled a graph of basically every number I could find.

This was possible due to two sources: “Swimming in small laboratory animals” Dawson and Horvath, 1970; and “Invasion ecology of Norway rats on New Zealand Islands” Russell, 2007. Both of these papers had collated lists of papers, which I happily pinched wholesale.

Let’s look at ONLY the red squares — which are the figures from Dawson and Horvath (1970) — and the black dots — which are the figures from Russell (2007). Each dot represents a group of normally-fed adult rats who swum at a temperature (x axis) for a time (y axis) from some study over the last 80 years or so. The overlap you can see between some of the points are due to my crap estimation skills, I had to copy the red points off a graph.

What you can see is that there’s a serious increase in ability to swim as the water temperature goes up. In fact, rats appear to be able to swim almost indefinitely in warm water... OK, hold that thought, we’ll get back to that part.

There were three groups in the original experiment:

  1. A diet of sucrose, casein, cottonseed oil and salt, with some added vitamins — A, B(1,2,3,5,6,7), C, D, E, K, folic acid. This group on the graph is marked STD.
  2. Same as above, but with a few extra fancy things added — biotin,PABA, inositol,B12. This is group STD-B.
  3. Same as 1. but with 10% dessicated liver powder. Group LIVER.

These groups (n=60; 3 x 20 each) were fed the above diets from about 3 weeks old (this is when they’re weaned) until they were 15 weeks old.

Now, the swim test. At 36*C, all these rats could swim comfortably for 120 minutes. That’s the blue circle you can see in the graph — all the rats are in the same place. Nothing else to say about that.

But at 20*C, you can see our ‘anti-fatigue’ factor at work… the green dots mark a few rats who didn’t make it to 120 minutes, and a lot who did. They did much better than the dark green circle (STD; standard diet) and the open square (STD; standard + extra B-vitamins).

Remember, the graph on the left is a LOG axis. The difference might not look big, but it’s a 4 to 9 times improvement.

This is where everyone stopped reading, which is unfortunate because it’s the time they should have gotten suspicious. That difference is huge. What’s going on?

The first puzzle: there are a fourth group of rats who are mentioned in the next section… as a comparison, the author tried rats on the same swimming procedure but with standard laboratory chow. They are shown on the my graph above as the orange circle.

Interestingly, under exactly the same conditions, this group did worse than the liver group. But they also did twice as well as the two STD groups.

What that implies raises another possibility — not the fact that liver is improving performance over baseline, but that liver might actually be correcting a deficiency in the control group diets.

The weight of the rats suggests that’s the case, as our liver eatin’ rats are in much better shape.

Data from Erschoff, 1951.

I did a quick one-way ANOVA on this — it’s significant. Then some t-tests, which are on the graph above. There is a negligible difference in weight between the liver and the STD+B group, and a MASSIVE difference in weight between the liver and STD group.

Alright, enough games. What’s the answer?

It’s iron.

The control rats are probably anaemic.

1. They have no dietary source of iron. The liver group, of course, has lots of it.

10% of their diet by weight was liver, no less. That’s plenty.

2. They DO have a source of Vitamin A, which is crucial for using the iron that they do have.

Buuuuuuut Vitamin A in the absence of iron doesn’t do much to help… and in any case, they still lose the head-to-head comparison between groups, as liver just happens to be stuffed with Vitamin A as well.

3. One of the symptoms of anaemia is impaired thermoregulation — the loss of the ability to successfully regulate body temperature.

“A number of investigations conclude that anemia is a central component of the inability of iron-deficient individuals to temperature regulate when they are cold stressed.”
Brigham and Beard, 1996

There’s actually a direct measurement of this in the study — our poor anaemic ratties lost body temperature a lot faster than their liver-eating friends.

… the body temperature of rats rapidly fell from 37*-37.5*C to 23*-25*C. The rate of fall was approximately 1*C per minute for the rats fed the basal ration … (and) about 0.6*-0.7*C for the rats in the liver group.
Erschoff, 1951.

Scroll back up and look at our green dots versus the rest of our dots — they’re actually on a very sensitive part of the graph. One thing you’ll notice is the massive increase in survival time from 20 to 25*C. In fact, at 24*C it’s about 3 hours, and 25*C it’s about 4 hours. It’s a place on the curve where more residual body heat/better thermoregulation could make a big difference.

4. Casein, their sole dietary protein source, interferes to a substantial degree with iron absorption.

Cow milk and cow milk products are an essential part of infant diet, yet they present a poor Fe absorption and so could contribute to the high prevalence of Fe deficiency in infants around the world; the affinity of cow milk proteins for Fe is so strong that it cannot be released and absorbed in a free form.
Kibangou (2005)

OK, that’s all the evidence in. Let’s look at the outcome — what are the consequences of Norwegian Rat anaemia?

Well, in the exercise physiologic sense, much the same thing as human anaemia — a lot less haemoglobin, which means impaired oxygen carrying capacity, which means your aerobic performance is in the tank.

Whole body VO2max was decreased by 18% by anemia regardless of training condition. Anemia significantly reduced endurance by 78% in untrained rats but only 39% in trained animals.
Gregg (1989)

Ever hear a vegetarian or a vegan admit (sometimes guiltily) that they ate a steak and felt much better? That’s haem iron.

The final nail in the coffin here is the realisation that this isn’t a supplement study. The control rats spent their whole lives with an iron deficiency. This is comparing a complete lifetime diet (or at least, mostly complete) with an incomplete lifetime diet. It doesn’t mean that liver has anything necessarily good or bad, probably just that liver improves symptoms of anaemia — no anti-fatigue factor required.

A point of interest — it’s interesting that the liver group still did better than the standard chow group, didn’t they? Would liver supplementation represent an improvement over standard chow? Potentially. We can’t really say what that means, though… rats can get awfully fat on lab chow. Also, I’ve looked high and low, and there’s no easily available record of what was in lab chow in 1951. Is there another deficiency to correct there? Impossible to say.

What should be obvious by now, at least, is that liver supplements claiming to be full of mysterious fatigue-fighting factors are bullshit. Especially those that glowingly cite this study. The study’s conclusion, rather than ‘adult men should take supplementary liver to achieve extra energy’ is ‘rats supplemented with liver through their entire lives outperform chronically anaemic rats’. Really isn’t the same thing, is it.

In closing, please keep in mind this isn’t a criticism of the original study. Science is littered with work like this, old studies which overlooked something that wasn’t well known at the time and end up looking silly in the cold light of history. Everyone’s science ends up looking silly eventually. This paper did. My work will too, some day.

And throw those liver tablets away, will you?