The Chocolate Life

Discover Chocolate and Live La Vida Cocoa!

Back in 2005 during my second trip to Ecuador to attend the University of Chocolate, our handbook contained a "formula" for how chocolate gets its flavor.

The formula included variables for the inherent flavors of cocoa beans and ways to think about the effects of fermentation, drying, and roasting on the final flavor of the chocolate. Because of this formula and conversations with chocolate makers I had come to believe what people told me about the need for fermentation for flavor development - which is to say that fermentation is absolutely necessary for any chocolate flavor to fully develop.

I learned on my recent trip to Belize during our time with the family of farmer Eladio Pop that that assessment is not entirely accurate. Here's why, using the framework of the formula for cocoa flavor development.

Cacao beans have inherent flavors. There are two "types" of inherent flavors, stable and unstable. Stable flavors are those that do not change during fermentation, drying, and roasting. Unstable flavors are those that are changed.

If you take cacao beans, wash the pulp off, and dry them, you do not subject the beans to the effects of fermentation and drying so both the stable and unstable flavors remain in the bean more or less unchanged. The stable flavors remain present through roasting and the unstable flavors change in ways that are different than if they had been fermented and dried. There are strong chocolate flavors in roasted unfermented cocoa beans but the range of flavors is pretty much limited to chocolate, roasty notes, nutty notes, and earthy notes.

During the fermentation process, sugars in the pulp surrounding the beans are first converted to alcohol and then to acetic acid. Proteins in the beans are broken down into amino acids, complex sugars in the beans are broken down to simpler sugars, and a wide variety of not-very-well understood chemical and thermal changes occur in the beans that create chemicals that contribute to the flavors found in fermented beans.

It is these new chemical elements that are created through fermentation that are responsible for the other flavors (e.g., red fruit, dried fruit) that show up in the finished chocolate. The availability of more sugars for the roasting process enhances sensations of sweetness and the chocolate, roasty, and nutty notes. The chemical and thermal destruction of the polyphenol compounds responsible for cacao's inherent bitterness contributes to these other flavors becoming more present through the effects of simultaneous contrast as they apply to taste. (Normally the principle of simultaneous contrast is applied to color in the fine arts.)

My epiphany was that fermentation is not necessary for the development of chocolate flavor in chocolate, it's necessary for the development of all the other flavors and for the development of more finely nuanced flavors. The demonstration of a traditional Mayan technique for making cocoa water was what showed me this. Farmer Eladio Pop does not ferment the beans his family uses to make cocoa water because it's too much work. Instead, the beans are washed free of pulp and dried. These beans are then roasted on a comal (a traditional flat metal griddle) over an open wood fire, cracked, winnowed, and then ground (using a hand corn mill) into a paste with black pepper and other spices (especially local allspice).

The mass that will not be used right away is rolled into balls and wrapped. The mass that is going to be used is mixed with water to form a more malleable paste that is then mixed with cold water and sugar to make the cocoa water. Even without fermentation the cocoa water had a very strong chocolate taste with the flavors of roasting and a strong nut (reminiscent to me of cashews) flavor. What was missing, I realized, was any trace of any of the other flavors that can be found in chocolate and it was at that point that I realized the real contribution that fermentation makes to flavor development.

It also explains, to a great extent, why cacao varieties that do not require much fermentation have flavor profiles in which nutty roasty flavors often predominate.

For me, it was a rather humbling realization as it became clear to me that I did not understand something I thought I understood very well. It also reinforced my opinion that travel to cacao-growing regions is absolutely necessary to develop a well-rounded understanding of cacao, cocoa, and chocolate.

Tags: fermentation, flavor development

Views: 1275

Replies to This Discussion

Clay- Very interesting post. It made me think about what I thought I understood about fermentation too. I still feel like, yes this is doable. You could end up with an edible product, but I imagine there are alot of things that could be done by skipping steps. You'd end up with an inferior product in most cases though, as compared to best practice. In the case of cacao, I thought without fermentation you end up with a sour, bitter, and astringent end product? Did you experience this in the drink you had? Do you think this practice was developed out of laziness or need for efficiency by the individual farmer you met or is this a widespread cultural practice? Since grading is based in part on fermentation, this would presumbably reduce the worth of the beans if it caught on. I'm still trying to grasp the idea of stable and unstable flavors. I wonder how true that is. Brady
Brady: Ultimately, I think that what happens is that different cultures use cacao for different purposes - and value different characteristics.

Historically, cacao pulp was used to ferment alcoholic beverages in South America; in Mesoamerica the seeds were used to create non-alcoholic beverages. Why? No one knows. How it occurred? No one knows. I think it may have something to do with domestication; in South America the trees were not domesticated and in Mesoamerica they were. It's a chicken and egg argument which came first (fermentation or domestication) but it makes sense to hypothesize that the two are related somehow.

I, too, thought that fermentation was required to eliminate (or at least reduce) bitter, astringent, and sour flavors. In this one experience, that was not the case. The beverage (cacau, or cocoa water) had three ingredients: roasted unfermented cocoa beans, sugar, and water. As I mentioned in the original post, the chocolate flavor was distinct with roasty and nutty notes. There was enough sugar to overcome any bitterness and there was no astringency or sourness. Also, there was no trace of what I now think of as fermentation flavors (e.g., fruits) and I think (now) that zero fermentation is probably preferable to partial (under) fermentation.

I think that perhaps the sour and astringent notes might have something to do with even partial fermentation. These seeds were taking out of the pods within hours of harvesting and washed more or less immediately - then dried.

I know that not fermenting beans for cocoa water it is a widespread practice among the Kekchi Maya of Southern Belize. I can't say for Mopan Maya or for other Maya tribes. I am reading a fascinating book called Household Ecology that talks about the way agriculture is weaved into the cultural fabric of Kekchi Mayan village and household life. Based on my reading of the book the Kekchi Maya are very sophisticated and can adapt a cash crop (e.g., cacao) to the needs of a cash market and make very sophisticated use and risk/reward calculations.

There is no way I can look at farmer Eladio and call him lazy. I think, however, that he has made a value judgment that says that the return for the extra effort of fermentation is not worth it for his own consumption. However, he is willing to undertake the work when he sells his cacao to the market for cash because he understands that the market values fermented beans. This is all tied into a complicated system tied to labor availability and other factors that non-Kekchi Mayua may never really comprehend.

With respect to stable v unstable flavors I am at a loss at the moment to come up with an analogy that is not entirely self-referential. I'll work on it.
Good points about the natural nuttiness of cacao. Now that I think about it, though, cacao is still a seed and like most seeds, should inherently have a nutty quality regardless of the presence of other chemicals and acids. The extent to which it's "nuttiness" is prominent depends on roasting, i.e. cooking, which as we all know changes and sometimes enhances flavors.

I've always wondered to what degree a chocolate's flavor is affected by differences in fermentation. Regardless of crop, every farmer has his own way of doing things, and I find it hard to believe that--despite what the experts say--there is a standardized procedure for fermenting. I think part of the problem is that we tend to get caught up in the romantic scientific facets (of how things should be done) and forget about the mercurial reality of farming.

In this respect, it seems to me to be very unreasonable to promote a vintage chocolate under the pretense that a harvest and environmental growing conditions are the sole contributors to a chocolate's flavor when in reality so many other factors are involved, namely fermentation, that also can affect flavor in dramatic ways. Perhaps, two batches of cacao from the same growing region that are fermented differently could conceivably produce entirely different chocolates after processing in the factory.

I don't know who is saying that there is a standardized procedure for fermenting. There are none (at least none that I have seen in my travels in Ecuador, Mexico, Costa Rica, Venezuela, and Belize - maybe there is on more "modern" farms in Ghana or Sao Tome or elsewhere).

You are correct in thinking that batches of cacao from the same growing region that are fermented differently can produce entirely different chocolates after processing. I know that Shawn Askinosie did some trials with the farmers who produce his Soconusco beans and at one point I had samples at three and four days fermentation that were very different smelling and looking at the inside. They would have had to have made very different tasting chocolate.

There might be recommended guidelines but there are too many variables in the process for fermentation ever to be consistent at any level. Remember these farmers have no thermometers, no control over the yeasts that innoculate the pile or over the range between daytime and nighttime temperature extremes or humidity levels. Some will ferment less than optimum, others will ferment more than optimum and it will be different for each pile they ferment.

In the end it all gets mixed together and what we end up with is an average that probably remains pretty constant.
Clay- It's very interesting to read about modern day uses of cacao in the regions in which they grow and then consider it in comparison to how we use it. When I read your first post I was inclined to assume it was a widespread cultural practice (as you said it is with the Kekchi Maya of southern Belize). Thanks for that follow up post.

I don't think you need to come up with an analogy for the stable/unstable flavors. I definitely understand what you mean. It's more the idea that cacao has stable flavors, not that I didn't understand what the description of a stable flavor is. Having not heard that before I was still unwilling to accept it at the time I read it. I'm warming up to the idea though and I don't see why that couldn't be possible. My original thinking was that a heavy roast might change a stable flavor when, considering this framework, it would more likely mask it.

Also, I'm not sure where I took this from, but in some old notes I wrote that unfermented beans would be less acidic, lighter in color, produce a less chocolate like flavor and have more of a green woody, bitter profile.
People often wonder where the flavors in chocolate come from. As Sam mentions, here are the chemical names of some chemical compounds (aldehydes and esters) that are responsible for chocolatey aromas/flavors in chocolate - and other foods.

2-methylpropanal (isobutyraldehyde),
3-methyl butanal (isovaleraldehyde) and
2-methyl butanal

If you want to learn more about these compounds (including the chemical names of more flavors and aromas) here are some resources to take a look at:

very from my own personal experience growing cacao and processing it, fermentation is absolutely necessary, and it has be fermented correctly, this can only be accomplished through trial and error for their are many factors involved. Unfermented beans made into chocolate is incredibly nasty, very very bad, even a nice criollo not properly fermented is horrible when made into chocolate. In my opinion as a cacao farmer on Kauai and chocolate maker who personally does all aspects of the chocolate art, fermentation is the most crucial step in the chocolate making process.

There is general forum thread on this topic that is very in-depth. Thanks to Sam Madell for her contributions to this conversation.

[Moderator's Note: - the link above no longer works. Sam left TheChocolateLife and removed all of the content she contributed when she left.]

For the truly obsessed:

2-Methylpyrazine ...C5H6N2
Odor Detection Threshold (in water) = 60,000 ppb
Green, nutty, cocoa, musty, potato, fishy-ammoniacal notes

Odor Detection Threshold (in water) = 6,000 ppb
Musty, nutty, buttery, peanut odor; chocolate-peanut taste

Odor Detection Threshold (in water) = 2,500 ppb
Green, nutty, potato, cocoa, coffee, caramel, meaty notes

Odor Detection Threshold (in water) = 800 ppb
Chocolate, roasted nuts, earthy; chocolate taste

Odor Detection Threshold (in water) = 200 ppb
Chocolate, roasted nuts, fried potato odor

Odor Detection Threshold (in water) = 400 ppb
Nutty, baked potato, roasted peanut, cocoa, burnt notes

Odor Detection Threshold (in water) = 1000 ppb
Weak, nutty, musty, chocolate odor; chocolate taste

Flavor Detection Threshold (in water) = 0.4 ppb
Potato, burnt nutty, roasted, cereal, earthy

Odor Detection Threshold (in water) = 100 ppb
Nutty, roasted, somewhat "grassy"

Odor Detection Threshold (in water) = 1 ppb
Cocoa, chocolate, nutty (burnt almond) notes

Flavor Detection Threshold (in water) = 0.4 ppb
Cocoa, chocolate, nutty (burnt almond, filbert-hazelnut) notes

Flavor Detection Threshold (in water) = 35 ppb
Powerful herbaceous green-earthy notes

Odor Detection Threshold (in water) = 400 ppb
Nutty, sweet, cocoa odor

Odor Detection Threshold (in water) = 3 ppb
Odor reminiscent of roasted peanuts, hazelnuts, almond

Odor Detection Threshold (in water) = 0.4 ppb
Roasted nut character; hazelnut, earthy

Odor Detection Threshold (in water) = 10 ppb
Earthy, bell pepper; vegetable; potato-like

Odor Detection Threshold (in water) = 0.002 ppb
Characteristic Green Bell Pepper aroma and taste in dilution

Odor Detection Threshold (in water) = 0.001 ppb
Green pea, green bell pepper-like odor and taste


Several closely related GRAS (generally recognized as safe) chemical structure types are responsible for providing the "burnt sugar" type notes associated with many products (caramel, cotton candy, maple sugar, cooked fruits such as strawberry & pineapple, as well as roasted products such as chicory & coffee). Here we will examine the flavor properties (odor descriptions and intensities) and show the molecular structural similarities.

As will be noted, each of these very important flavor chemicals possesses the alpha-enol function adjacent to a carbonyl. Also of interest is that the replacement of methyl groups with ethyl groups in these structurally similar compounds generally results in an amplification of odor intensity.

Maltol or 3-hydroxy-2-methyl-4-pyrone...C7H8O3
is an important flavor chemical that occurs naturally in products such as Caramel, Chicory, Cocoa, Coffee, Milk, Roasted Malt, Strawberry, and Bread ... just to name a few. The odor is often described as that of "cotton candy", the spun caramelized sugar product sold at fairs.
Odor Detection Threshold (in water) = 35,000 ppb
Sweet, fruity, berry, caramellic odor; fruity preserve-like. Very important in commercial fruit flavors.

Ethyl Maltol or 3-hydroxy-2-ethyl-4-pyrone...C8H10O3
is the ethyl analog of Maltol, wherein the 2-methyl group is now 2-ethyl. This material is not occurring in nature. The odor is also that of "cotton candy"; in fact the odor/flavor properties are nearly identical except that Ethyl maltol is 4-5 times as intense in flavor applications and perhaps slightly more fruity. I have been unable to find a published value for its odor detection threshold, but would expect it to be in the range of 10,000 ppb.
Odor Detection Threshold (in water) = NA
Sweet, fruity-caramellic odor; fruity preserve taste

Furaneol(R) or 2,5-Dimethyl-4-hydroxy-3(2H)furanone ..C6H8O3
also known as "Strawberry furanone" or "Pineapple furanone" was originally reported in 1965 by Willhalm, Stoll & Thomas at Firmenich & Cie in Geneva as a key flavor component of strawberry [Chemistry & Ind. [London], 1629, (1965)]. At essentially the same time, Rodin, Himel, Silverstein, Leeper & Gortner [J. Food Sci., 30, 280 (1965)] found this material to be one of the main organoleptic principles of pineapple. Today, we know that Furaneol plays an important roll in the flavor of numerous fruits (Chempedak Fruit, Guava, Litchi, Pineapple, Raspberry, Arctic Bramble, Strawberry, Tomato and others), as well as in roasted products such as coffee, corn tacos (maize), cooked beef, malt, hazelnut, roasted almonds, and popcorn. While the odor threshold has been reported by various workers at values ranging from 0.4-1700 ppb, the value of 31 ppb seems most reliable. Buttery has shown that the threshold value in water is pH dependant. Odor threshold in water: 60 ppb at pH 7, 31 ppb at pH 4.5 and 21 ppb at pH 3.0 [Buttery, Ron G.; Takeoka, Gary R.; Ling, Louisa C., J. Agric. Food Chem., 43(6), 1638-40 (1995)]. Furaneol is a trademark of Firmenich.
Odor Detection Threshold (in water) = 31 ppb
Fruity, caramelized pineapple-strawberry odor & taste

Furaneol acetate or 2,5-Dimethyl-4-acetoxy-3(2H)furanone ..C8H10O4.
is the enol acetate of Furaneol. This material was added to the FEMA (Flavor extract Manufacturers Assn.) GRAS list in 1996. The author has been unable to find a reference either to the occurrence of this in nature or an odor detection threshold. However, we find this material to possess better stability to air than furaneol and believe it to have a superior cooked strawberry note. It probably finds applications wherever furaneol can be used.
Odor Detection Threshold (in water) = NA
Cooked caramelic strawberry note

Mesifurane or 2,5-Dimethyl-4-methoxy-3(2H)furanone ..C7H10O3.
Although this material accompanies the related furaneol [2,5-dimethyl-4-hydroxy-3(2H)-furanone] (FEMA# 3174) in Pineapple, Raspberry, Strawberry and Grape (and in fact is higher in concentration in some cases, such as Arctic Bamble) - and the material has a marginally lower odor threshold. This material does not excite the flavorist as much because its flavor is much more subtle and mellow. This material is an excellent blender in Fruit flavors.
Odor Detection Threshold (in water) = 0.3 ppb
Sweet, fruity "sherry" like odor; Xeres wine-like note

Cyclotene or 3-Methyl-2-cyclopenten-2-ol-l-one ..C6H8O2
resembles furaneol and maltol in that it has a 5-membered ring with its enol carbonyl structure. Occurs in Cocoa, Coffee, Fenugreek, Licorice, Malt, Roasted Almond and almost all products containing sugar that are roasted. In conjunction with fenugreek solid extract, "cyclotene" is an integral part of nearly all Maple flavors. Several authors report the flavor of this material to be somewhat similar to Licorice but, to most North Americans, its flavor is closer to the syrup obtained from a sugar maple.
Odor Detection Threshold (in water) = 300 ppb
Very strong, caramellic-maple, lovage odor and taste

3-Ethyl-2-hydroxy-2-cyclopenten-1-one or Ethyl cyclotene ..C7H10O2
is the ethyl homolog of cyclotene and possesses a similar odor and taste. Occurs in Coffee, Tobacco & Tobacco smoke. Fenaroli (1975) indicates the flavor threshold value for this material is "5 ppm" and that it also acts as a flavor enhancer. On a relative basis to cyclotene, we expect that a more accurate threshold value is about 150 ppb (however, this is speculative, at this point).
Odor Detection Threshold (in water) = 5000 ppb
Very strong, caramellic-maple odor and taste

3,5-Dimethyl-2-hydroxy-2-cyclopenten-1-one or Coronol(R) ..C7H10O2
is 3,5-dimethyl analog of cyclotene and possesses a similar odor and taste. Occurs in Coffee & Tobacco smoke. Similar strong maple-caramel notes. Coronol is a trademark of Givaudan-Roure.
Odor Detection Threshold (in water) = 1000 ppb
Strong, maple-caramel, nutty odor/taste; similar to cyclotene

3,4-Dimethyl-2-hydroxy-2-cyclopenten-1-one or Methyl Corylone(R) ..C7H10O2
is the 3,4-dimethyl analog of cyclotene and possesses a similar odor and taste. Occurs in Coffee, Tobacco Smoke & Wood Smoke. Similar strong maple, burnt-sugar, caramel odor/taste. This material is the most potent of the cyclotene analogs. Methyl Corylone is a trademark of Givaudan-Roure.
Odor Detection Threshold (in water) = 17-20 ppb
Strong, maple-caramel, nutty odor/taste; similar to cyclotene

3-Ethyl-2-hydroxy-4-methylcyclopent-2-en-1-one ..C8H12O2
is the ethyl homolog of 3,4-Dimethyl-2-hydroxy-2-cyclopenten-1-one and has a somewhat more burnt, caramellic, maple-like odor. To the authors knowledge this material has only been reported in Tobacco smoke. Similar strong maple-caramel notes.
Odor Detection Threshold (in water) = NA
Strong, burnt, caramellic, maple-like odor

3-Ethyl-2-hydroxy-5-methylcyclopent-2-en-1-one ..C8H12O2
is the ethyl homolog of 3,5-Dimethyl-2-hydroxy-2-cyclopenten-1-one and has a more burnt, caramellic, maple-like odor. To the authors knowledge this material has only been reported in Tobacco & Tobacco smoke. Similar Burnt, caramellic, maple odor as the 3-ethyl-4-methyl isomer.
Odor Detection Threshold (in water) = NA
Strong, burnt, caramellic, maple odor

Sotolon or 4,5-Dimethyl-3-hydroxy-2(5H)-furanone..C6H8O3
also known as "Caramel furanone", is the key organoleptic principal of roasted Fenugreek seed. Also found in Roasted Virginia Tobacco, Sake, Sugar, Botrytized grape wine and Rice wine. Until the materials shown on this page became available, Fenugreek and Fenugreek extracts were the only good flavor concentrate source for caramel-maple notes. A product known for many years under the trade name of "Mapeline" was used by housewives and industry whenever this type of flavor was needed. Perfumers valued the powerful Fenugreek absolute for such notes. The taste and aroma of brown sugar is mainly due to sotolon (which is the main flavor principal in sugar molasses). However, Sotolon is now available commercially and is one of two extremely powerful aroma chemicals for imparting caramel-maple notes. Sotolon with a detection threshold of 0.001 ppb is nearly 30,000 time more powerful than cyclotene. The chemical which follows is even more powerful.
Odor Detection Threshold (in water) = 0.001 ppb
Powerful caramel aroma; sweet burnt taste

Maple furanone or 5-Ethyl-3-hydroxy-4-methyl-2(5H)-furanone..C7H10O3
is truly one of the most outstanding of modern flavor materials. This material is much more powerful and lasting than Methyl or Ethylcyclopentenolone and possesses more of a maple note than Sotolon. Maple furanone is a key organoleptic note in Soy sauce (hydrolyzed vegetable protein). With a detection threshold of 0.00001 ppb maple furanone is nearly 3,000,000 times more powerful than cyclotene and in fact is one of the most powerful flavor chemicals known to man.
Odor Detection Threshold (in water) = 0.00001 ppb
Powerful maple-caramel aroma and taste
Dear NerdZone friends:

I was reading all your inputs and comment (accept the long list of flavor precursors) and how I see it, it is a very sophisticated exchange of knowledge and high level discussion. You should know that I came a long way to specialty cacao fermentation, being an agriculturist working with smallholders almost 20 years.

Clay mentioned, there are traditional “standard procedures” the people in a certain region follow more or less and the mixture of the cacao normally comes out the same way every year. Some regions are better than others, what has also to do with the specific climate, soil, varieties and the mix of it.

Meanwhile I did many tests with fermentation also using different techniques and it is true, the chocolate comes out different every time one changes an element. For smaller specialty batches and estate cacao one can create "typical flavor pattern", but it is too complicated for smallholders and it needs specialty markets to pay off the efforts.

The level you are talking one has to look at fermentation more like a wine maker and start to "industrialize" the whole fermentation and drying process to be sure to get the same result every time. We also cannot leave the fermentation process to spontaneous colonization of some unknown yeast strains. We have to control temperature, moisture and mixing etc. That is not knew. I found technical books from the 60ies with design for cacao fabrics for Africa. It was dropped somehow and I presume that it was too costly and there was no demand for such a sophisticated cacao at the time.

I heard that Nestle recently told the Ministry of Cacao in Ghana not to (improve) change the (traditional) fermentation pattern, as they will not accept a different taste. (I think that Nestle is not willing to pay more for the cacao to the farmers for more work during the fermentation process.)

The range of cacao qualities lies now between:
1. The wisdom and tradition of cacao producers and
2. development of new fully controlled standards.

We know pretty much how farmers in certain areas work and that changes in “quality” can be made with minor technical improvement if wanted. Controlled industrialized procedures are not well developed. On mayor problem will be to control the micro organism involvement. Recent studies had a look at the different yeasts strains, which are "typical" for each region. They looked also at the involvement and timing of their appearance also of acid bacteria and lacto bacter, and their possible influence to flavors and the changes if mixture and timing.

Maybe someday we will have “home-fermentation-units” with a toolbox of different starter cultures, which can trigger certain flavor developments!? (Be patient I am working on it.)
great ideas about flavor developing processes in cocoa. It´s is refreshing to know that fermentation is not the only way t get flavor and aroma out from the beans. I remember now a Dominican varietal that is sold internationally that is not not fermented at all, as a traditional way of treating this type of cocoa. I do not know the particular reason behind in this case.
Since you have been around you might know about¨arriba beans from my country, do you have some specific tips about arriba fermentation specifications¨?
I´ve read your conversation with great interest, and, well, since I just today felt my preconceptions on cacao processing quite "shattered" I would like to hear your opinion on the following...
Investigating on machinery to process cacao beans into chocolate, I was told that roasting wasn´t necessary to achieve a good development of flavors in chocolate if the beans had been properly fermented and homogenized (especially in terms of their moisture content) before.
[that the optimal time of fermentation for any kind of cacao bean was 3 days, that anything else would only decrease the bitter flavor notes of cacao an increase a bitterness which comes from tannins contained in the pulp...?]
further, that some kind of centrifuge could replace a conche in order to finally obtain the coberture.
I never heard about this before. well, I´m not claiming to know it all, but still, I´m wondering...


Member Marketplace

Promote TheChocolateLife

Bookmark and Share

Follow Clay on:
Twitter :: @DiscoverChoc
F'Book :: TheChocolateLife
F'Book Group :: LaVidaCocoa :: @DiscoverChoc



© 2014   Created by Clay Gordon.

Badges  |  Report an Issue  |  Terms of Service