Caffeine

[Instaurator]

There was an article published recently about caffeine and it perpetuated some urban myths that I have heard one time too many and so Alistair kindly invited me to let off a bit of steam and to hopefully help some people who must be very confused about caffeine at least.

In this article it said: "Another misconception about the amount of caffeine in coffee and espresso is that the darker the coffee, the more buzz it must have�.

This is actually not a misconception. It is in fact true that darker roast coffee does have a higher percentage of caffeine.

As a percentage of 'dry matter� Arabica green has 1.2% whereas roasted it has 1.3%. Robusta green has 2.2% and roasted it has 2.4%. [Ref Dr Illy]

What happens during roasting is that there is a very small amount of caffeine loss, only a few percent, whereas a medium roast will lose 5-8% organic matter and a dark roast will lose 8-12% organic matter. Very dark roasts will lose more than 12% organic matter.

So that is why in absolute terms there is actually a loss of caffeine (which is what the article may have been referring to) but in relative or percentage terms there is more caffeine, the darker the roast. This is simply because organic matter and moisture is lost at a greater rate than caffeine.

So in 10grams of dark roast coffee for an espresso will have more caffeine than 10 grams of lighter roast. It becomes more not less concentrated.

I hope this makes sense.

This particular article said: 'Many people " customers and baristas alike " are surprised by this fact [that dark roasts are not meant to give you more 'buzz'] at first...�. These people should be surprised for a very good reason: their bodies are confirming the truth that they are experiencing more caffeine and more 'buzz' and unfortunately if they have been told otherwise it is not a 'fact�. It is another coffee myth or 'DFO' (Data Free Observation) as my agronomist would call it.

And what is more, caffeine doesn't seem to translate to the cup in a linear fashion, not for espresso anyway. It is often said that robusta has twice as much caffeine than Arabica.

A pure Arabica blend for an espresso (30mls in 30 seconds) will contain 2.6mg/ml whereas a pure robusta blend will contain 3.8 mg/ml. This is actually not twice as much but just under half as much again.

Sorry to bore everyone but at least I've got it off my chest now .

[Jeff Givens]

That's interesting stuff. As I understand from your posting, by weight, the relative percentage of caffeine increases with darker roasts; but by volume, the percentage decreases.

If my interpretation is correct, wouldn't a darker roast have a greater percentage of caffeine if the barista is dosing into the basket by volume?

[Matthew Brinski]

That is not boring at all ... thanks for sharing.

(I hope you let off some more steam.)

[Instaurator]

"As I understand from your posting, by weight, the relative percentage of caffeine increases with darker roasts; but by volume, the percentage decreases. "

Yeah that's another way of expressing it - a little more clearly too! So you have less caffeine matter but at a higher percentage or concentration.

Sure if you dose more, automatically you have more of everything! But I wasn't thinking in terms of dosing or 'updosing' as some people like to call it. By the way I'm not hung up at all on that as being the only way of dosing. For what it is worth there is a whole other theory on dosing that maybe I'll post some time. Actually it's not mine but Scotty Callaghan's the current Aussie Barista Champ.

[barry]

So that is why in absolute terms there is actually a loss of caffeine (which is what the article may have been referring to) but in relative or percentage terms there is more caffeine, the darker the roast. This is simply because organic matter and moisture is lost at a greater rate than caffeine.

I usually describe this as "bean for bean, less caffeine; pound for pound, more caffeine".

I distinctly recall answering this question correctly during a trivia-for-tshirts interlude during the barista competition at the san francisco convention (standing next to jeff taylor). I was promptly chastised by the "expert" and several members of the audience who, quite obviously, hadn't read their illy. I did not get the t-shirt, which is okay, because we dropped Torani shortly thereafter.

[barry]

"As I understand from your posting, by weight, the relative percentage of caffeine increases with darker roasts; but by volume, the percentage decreases. "

Yeah that's another way of expressing it - a little more clearly too! So you have less caffeine matter but at a higher percentage or concentration.

I'm not sure that's accurate, because relative volumes depend upon bean density, not just bean mass. bean for bean, the amount of caffeine decreases.

[Instaurator]

Barry said:
"I'm not sure that's accurate, because relative volumes depend upon bean density, not just bean mass. bean for bean, the amount of caffeine decreases."

I'm not sure I understand what you mean. I missed out on a whole lot of physics at high school by spending way too much time at the beach surfing. But as I understand it, density is mass divided by volume. Whereas weight is mass multiplied by gravity. So 1 pound of coffee on planet earth will weigh 1/6 of a pound on the moon but it will have the same mass. Density doesn't change unless the volume of the same 1 pound changes. But this is all very abstract. The fact remains when we drink a cup of coffee made from the same mass of coffee, a dark roast will have a higher percentage of caffeine in comparison to a light roast of the same coffee. This is of course assuming we are not weighing and brewing our coffee on the moon....or some other planet. I'm not saying you are doing this Barry but sometimes I fear some people must be.

[Jim Saborio]

As a percentage of 'dry matter� Arabica green has 1.2% whereas roasted it has 1.3%.[Ref Dr Illy].

With this in mind, I'd imagine the difference between a light and dark roast would be even smaller. Would this small increase be significant to the consumer?

At my first barista gig, our roaster gave us a sheet detailing varying caffeine contents between origins. People really latched on to the "fact" that our Tanzanian peaberry had the most caffeine of our offerings. Some would avoid the Tanzanian like the plague... "Whoah, I don't want to be bouncing off the walls!"

I suspected the difference was largely in their heads.

Should we be telling folks "Yes, dark roasts generally DO contain more caffeine than lighter roasts BUT the difference is negligible"?

I'm confused.

[barry]

Density doesn't change unless the volume of the same 1 pound changes.

Exactly. The volume of a pound of dark roast coffee is different than the volume of a pound of light roast coffee. What the difference is depends upon bean, relative degree of roast, and roasting method, so making a statement on caffeine in a given volume of beans (which is {perhaps incorrectly?} how I read the initial volume comment) is difficult, at best.

The fact remains when we drink a cup of coffee made from the same mass of coffee, a dark roast will have a higher percentage of caffeine in comparison to a light roast of the same coffee.

Agreed wholeheartedly.

[barry]

Should we be telling folks "Yes, dark roasts generally DO contain more caffeine than lighter roasts BUT the difference is negligible"?

This one.


The difference depends entirely upon the relative percentage of caffeine at any given roast level. The "1.3% roasted" value is not exact, and ought to vary with roast level... in fact, it must vary with roast level for there to be a difference at all based on mass (I'll show the math if you want). In other words, 1.3% caffeine in a dark roast and 1.3% caffeine in a light roast result in the same 1.3g per 100g sample.

If you want to make up numbers for demonstration purposes, perhaps assume 1.29% for medium roast and 1.31% for dark roast, and there will be a .02% difference based on mass. I'll see if there are any references which show %caffeine based on roast degree.

BTW, it's easy to demonstrate that bean for bean, there is less caffeine. I'll show the math for that, too, if you want.

[Jim Saborio]

The fact remains when we drink a cup of coffee made from the same mass of coffee, a dark roast will have a higher percentage of caffeine in comparison to a light roast of the same coffee.

Agreed wholeheartedly.

As someone who might have 10 seconds to clear-up a consumer's caffeine misnomer, I'm more concerned about the results in the cup. That's what people care about. Pound for pound, bean for bean, or content by volume is just fun trivia that may or may not help you win a t-shirt.

Supposedly a 12oz can of Coke has 3.5mg more caffeine than a 12oz can of Pepsi.

Is this the kind of difference we're dealing with here?

I doubt there are many consumers who choose Coke because the extra .291 mg/ounce gives them the edge they need in today's hectic world. At the same time I wouldn't expect many to choose Pepsi because their delicate hearts can't handle Coke.

Even if there's a 25mg difference between two 16 oz. cups of coffee, is this something a consumer should concern themselves with?

Will they physiologically notice the difference or is it just fun trivia?

That's what I was attempting to ask.

[barry]

Should we be telling folks "Yes, dark roasts generally DO contain more caffeine than lighter roasts BUT the difference is negligible"?

Again, the above is the correct response (imho).

You're not going to get hard numbers, because caffeine is a natural component in coffee, and final content in the cup depends upon way more variables than you'd be able to account for precisely. It's easier with soda, because caffeine is an additive, and the manufacturers know just how much they put in, and lab tests on one can of Coke pretty reliably reflect the content in other cans of Coke. Not so with coffee. So, don't look for hard numbers, just look for general relationships of contributing factors. If you use 8 grams to brew a cup of coffee, and the bean has 1.3% caffeine, then the cup will have 104 milligrams of caffeine in it provided 100% of the caffeine is extracted. Not all brew methods achieve 100% extraction of caffeine, caffeine content in the bean varies somewhat, and dose values vary somewhat, so I would be reluctant to start throwing hard numbers at customers w/o some lab analysis.

[Brent]

I suspected the difference was largely in their heads.

I'm confused.

In a recent (last 2 years?) issue of National Geographic, they covered caffeine, and one of the points that came out was that caffeine does not affect the system immediately, rather about an hour later, hence that "instant kick" IS all in your mind...

[Andy Schecter]

In a recent (last 2 years?) issue of National Geographic, they covered caffeine, and one of the points that came out was that caffeine does not affect the system immediately, rather about an hour later, hence that "instant kick" IS all in your mind...

The article claims the effects don't PEAK for an hour.

Even if true, that's very different from saying there's NO effect for an hour.

[Brent]

The article claims the effects don't PEAK for an hour.

Even if true, that's very different from saying there's NO effect for an hour.

I was remembering the article, not quoting

[Instaurator]

As someone who might have 10 seconds to clear-up a consumer's caffeine misnomer, I'm more concerned about the results in the cup. That's what people care about.

Jim said
'Yes, dark roasts generally DO contain more caffeine than lighter roasts BUT the difference is negligible"?

I'm confused.�

The aim here was to hopefully try and dispel some confusion. Jim's point about whether it is noticeable in the cup is great! That is what counts. If something doesn't make a difference in the cup than the debate becomes as practical as a medieval discussion about how many angels there are on the point of a needle.

For what it is worth, a roast at Agtron whole-bean 60 (say light/medium drip-filter roast) will equate to about 80 beans per 10 grams versus a roast at Agtron whole-bean 13 (say Italian or French very dark roast, depending what part of the country you are in) will equate to 90 beans per 10 grams. This is 12% more per cup on top of the greater percentage that has already been agreed to.

Given that most of us either weigh the same amount of coffee for each brew of drip-filter or dose it to our own recommended level in a porta-filter, regardless of roast variations (Again assuming we are on planet earth, with who knows how many angels around!) I would suggest this is a substantial increase in caffeine not a 'negligible� increase, without wanting to get into semantics.

And given that everyone's caffeine tolerance varies much like alcohol tolerance, it is hard to say how it affects individuals physiologically, but I would venture to say it is noticeable by most people. No data here, but certainly if you have an empty stomach because you've missed breakfast and you don't have milk with your first coffee of the day it is pretty easy to feel that caffeine hit.

What Barry said is true in regard to quantifying the amount of caffeine. It is very difficult to quantify this in real-time due to the variability of how it is extracted; type of beans; how much the internal of the beans has been roasted; etc but in 10 seconds you can say to your customers: yeah in general a dark roast has more caffeine but also recommend they find their own enjoyment and tolerance level. At the very least they should definitely not be told there is less caffeine in a darker roast cup of coffee!

[Sean Starke]

Would it help to explain in terms of the serving size? It seems to me that's what the consumer cares about: how much caffeine is in my shot of espresso as compared to my 12 oz regular cup?

[Ric Rhinehart]

How many angels will fit on the head of a pin? This conversation is interesting but operating out of the realm of the real world. Please read this thread Starbucks vs Caffeinefor an interesting background discussion.

Consider the enormously broad standards we apply to coffee. A shot of espresso needs 7-9 grams of coffee, a range in excess of 20%. A gold cup brew requires 3.25 - 4.25 ounces of coffee for 1.9 liters of water, a range of more than 20%. Green arabica coffee has caffeine content by mass of 1.0 - 1.4%, again a range in excess of 20%. A shot of espresso ranges from 1 -1.5oz in volume, a range in excess of 30%. Roasted coffee is 28 - 35% soluble...you get the picture.

Caffeine is moderately soluble in water and is increasingly soluble with higher temperature. I would be willing to bet that solubility increases with pressure and with contact time as well, so throw in some more variables. All this without accounting for any operator errors. Complex question without an answer.

[Edwin Martinez]

This conversation is interesting but operating out of the realm of the real world.

What does the real world have to do with anything? (i'm teasing, ric)

I can appreciate the big picture and understand how context can turn a 180 on just about any hard fact statistic. But am glad to see coffeed is a place for the real world as well as the methodical mad scientist that continues to push the envelope.

I learned something new this morning.

[barry]

For what it is worth, a roast at Agtron whole-bean 60 (say light/medium drip-filter roast) will equate to about 80 beans per 10 grams versus a roast at Agtron whole-bean 13 (say Italian or French very dark roast, depending what part of the country you are in) will equate to 90 beans per 10 grams. This is 12% more per cup on top of the greater percentage that has already been agreed to.

Given that most of us either weigh the same amount of coffee for each brew of drip-filter or dose it to our own recommended level in a porta-filter, regardless of roast variations (Again assuming we are on planet earth, with who knows how many angels around!) I would suggest this is a substantial increase in caffeine not a 'negligible� increase, without wanting to get into semantics.

Except the data on caffeine content is based on mass of beans, so 1.3% caffeine, by weight, is 1.3% no matter what the density of the coffee. Bean count is irrelevant when dealing with mass. Sort of like the old joke, "which weighs more, a pound of feathers or a pound of gold?"

If you use 65g/l for drip brewing, then, ceteris paribus, the difference in caffeine content between two brews would be whatever difference in caffeine content there is between the two roasts, ie, most likely not significant, even though the light batch used 520 beans and the dark batch used 585 beans. In other words, the two brews made with 65g/l would each yield 845mg caffeine, even though the dark brew was made with 12.5% more beans, if the caffeine content is 1.3% by weight.

So, what we need to know is what the relative caffeine content is based upon roast color. IIRC, the first edition of Illy had a table with this data in it. Unfortunately, that copy of mine is out on loan and I can't seem to find the equivalent data in the new edition.

I don't think Sivetz addresses this, either, but I'll double-check to make sure. It's probably in Clarke, but I haven't that one.

[barry]

Okay, another way to look at it which might make things clearer (I hope).

Browsing through Coffee Technology finds the assertion, "Hardly any caffeine is lost in the roasting." (Sivetz, pg 562) The following example is also provided, "...if caffeine is 1.0 percent in green coffee and there is a 16 percent roast loss, then 1.2 percent caffeine will result in the roast coffee." (Sivetz, pg 562).

1% = 1g/100g sample. A 16% roast would yield 84g of coffee, with essentially that same 1g of caffeine. 1/84 = 1.19%

So, let's spread this out a bit further. The assumption is that the absolute value of caffeine remains the same regardless of roast loss (note it's the absolute value, not any percentage value).

1% = 1g/100g sample. A 15% roast yields 85g, for a final caffeine content of 1/85, or 1.17%.

A dark roast, of 20% weight loss, yields 80g with that same 1g of caffeine, or a final content of 1/80, or 1.25%.

So, there you go, a darker roast has more caffeine, by weight, than a lighter roast. In a 65g/l brew, ceteris paribus and assuming 100% extraction of caffeine, the medium roast would have 760.5mg of caffeine and the dark roast would have 812.5mg caffeine (or only 6.8% more caffeine).

In a 300ml cup, that would be 228mg for medium and 243mg for dark.

Much clearer that way.

[Instaurator]

Except the data on caffeine content is based on mass of beans, so 1.3% caffeine, by weight, is 1.3% no matter what the density of the coffee. Bean count is irrelevant when dealing with mass. Sort of like the old joke, "which weighs more, a pound of feathers or a pound of gold?"

Yeah but a pound of caffeine is going to have more caffeine than a pound of feathers or a pound of gold. And that was the point I was trying to make in regard to the darker the roast, the more concentrated the caffeine. It won't stay at 1.3% if you go darker still. This is supported by the 6.8% Barry came up with. If the 1.3% Illy figure came from a 'Normale' roast that would be about 47-50 wholebean Agtron. If you go right down to a Agton 13 you have to make up your target weight for your brewer with more beans. This results in more caffeine per cup.

[zak]

Let me pose this, which may be rather moot. Let's say you roast a pound of bean A to a let's say a city roast but there are far more beans in a pound because as a green bean A weighs less than bean B and lets say you take bean B to a city plus. If bean A at this point is now lighter but after weight loss during roast is equal to bean B so will now take an equal amount of beans to obtain it's respective brew parameter say 7-9 grams for an espresso. Wouldn't it now be true that the lighter roast now contains more caffeine.

Josh Longsdorf
Ugly Mug Cafe

[barry]

Let me pose this, which may be rather moot. Let's say you roast a pound of bean A to a let's say a city roast but there are far more beans in a pound because as a green bean A weighs less than bean B and lets say you take bean B to a city plus. If bean A at this point is now lighter but after weight loss during roast is equal to bean B so will now take an equal amount of beans to obtain it's respective brew parameter say 7-9 grams for an espresso. Wouldn't it now be true that the lighter roast now contains more caffeine.

That depends upon the initial caffeine content in the green, and the roast weight loss of each particular roast. You've posed a problem w/o sufficient data to support your conclusion.

Or, as I used to say in symbolic logic class, "If you're going to assume all that stuff in the beginning, why not just assume the answer and we can all go home early?"

[zak]

Or, as I used to say in symbolic logic class, "If you're going to assume all that stuff in the beginning, why not just assume the answer and we can all go home early?"

I think you've kind of gotten to the point I was trying to make in that it will always depend on more variables than a barista may know. Therefore we cannot conclusively state one or the other. Every coffee is going to be different. The only thing we can actually state is that if a coffee is roasted to two varying degrees, the darker will have more caffeine. There are many baristas out there and I would venture to say most, even of the most well trained cafes, that do not have the background knowledge to the extent they need to answer a question about two or three or however many coffees they may have available. So if we are looking at the question "which has more caffeine?" from the perspective of some one serving the customer, how now do we answer it?

Josh Longsdorf
Ugly Mug Cafe