There are two key texts preserving recipes for fish sauce. The most reliable evidence for the manufacture of these sauces is found, not in Latin sources, but in a 10th century Byzantine Greek agricultural manual, the Geoponica (Dalby 2011). Despite its medieval date, it actually preserves Greek material from much earlier in the Roman period and is therefore a valuable source. It is clear that, if these fish sauces were not Greek in origin, their use was at least spread by Greek cooks through their recipes. The definitions in the Geoponica are coherent and are not contradicted by any other didactic text. There are four separate methods to make various types of fish sauce.
The making of garaFootnote 5. The so-called liquamen is made thus. Fish entrails are put in a container and salted; and little fish, especially sand-smelt or small red mullet or mendole or anchovy, or any small enough, are all similarly salted; and left to pickle in the sun, stirring frequently. When the heat has pickled them, the garos is got from them thus: a deep close-woven basket is inserted into the centre of the vessel containing these fish, and the garos flows into the basket. This, then, is how the liquamen is obtained by filtering through the basket; the residue makes alix. (Geoponica 20.46; Dalby 2011).
The text goes on to describe a similar method but with no extra viscera, where small and larger fish, that we can assume were cut open, are used and the method of making garos is described whereby whole fish are boiled in brine. This is mirrored in a recipe from the obscure Brevorum of Ps Rufius Festus (Curtis 1991: 192; Eadie 1967) which is considered a Medieval gloss, as is the other recipe source, found in two of the manuscripts attributed to Gargilius Martialis’ Medicinae ex holeribus et pomis (Ps Gargilius 62; Rose 1875: 209) where a similar fermented sauce made in an enclosed vessel with many herbs is described (Maire 2002; Curtis 1984). The definition from Isidore of Seville’s 6th century encyclopaedia recalls these recipes: ‘liquamen is so called because little fish dissolved during salting produce the liquid of that name’ (20.3.19ff). That this dissolved liquamen sauce could also be made from much larger fish such as mackerel (Scomber sp.) is clear from the tituli picti where occasionally liquamen scombri flos occurs.Footnote 6 However the bone residues of fish sauce that are found throughout the Mediterranean at production and consumption sites are dominated by the small fish species such as sardine and anchovy (Engraulidae, Clupeidae) and it is these that are used for the mass produced liquamen (Desse-Berset and Desse 2000).
It is apparent from the first three instructions in the Geoponica and from the Gargilius technique that a sauce derived from whole dissolved small fish, whether with extra viscera or not, can be made by three methods: fermentation of the small fish (5–20 cm) with extra viscera and salt in bulk, open to the sun; fermentation of fish in an enclosed vessel, cut in pieces if large; making a sauce by boiling fish in brine. The result is the same. In my experiments to make fish sauces, all these methods have resulted in sauces of a similar nature, once the bones were removed: a light or darker brown clear liquid containing very small particles of the fish paste residue floating in it (Grainger 2016).Footnote 7
Later sources state that the term garos had a generic meaning in Greek for any salty liquid, even one extracted from meat, and in Latin a liquamen derived from salting and fermenting pears is found in Palladius.Footnote 8 We can see from this and from the Geoponica that a generic usage indicating a salty liquid seasoning works with garos and liquamen. The question is does it also work with garum? I would suggest not!
A very different fish sauce results from the fourth instruction in the Geoponica which constitutes the second type of sauce thus:
A rather high quality garos called haimation is made thus: Take tunny entrails with gills fluid and blood, sprinkle with sufficient salt, leave in a vessel for two months at the most; then pierce the jar and the garos called haimation flows out (26–29).
This blood/viscera sauce is a dark and rich and packs a powerful punch. It is iron rich and having made such a sauce I can confirm that it is very different in taste to any of the whole fish sauces. This is largely due to the blood which we learn has to be harvested while the fish is freshly killed otherwise the blood coagulates and is unobtainable (Van Neer and Parker 2008). Garos haimation is certainly sufficiently different from a whole fish sauce to justify the need for a clear distinction in the terminology essential in the manufacturing process and in terms of trade. The distinctiveness is also apparent when we see how these sauces were used (see below). This second type of sauce is also described as a garon melan (black) in Greek and variously rendered in Latin as garum/garum sociorum or nigrum (black). We do not find the word haimation (bloody) transliterated in Latin sources, it is not known on tituli picti and it is quite rare, with one known reference in Egyptian papyri.Footnote 9 The identification of the sauce known as garum sociorum is greatly disputed (Leon 2001: 176). It is not generally accepted that it is the same as the haimation sauce and it may represent another product although, as we shall see, the ancient writers identified a sauce made with blood with the label sociorum.
There is by no means universal agreement on this basic issue of the two types of sauce being distinguished in this way. The three different production methods, noted above, are viewed as producing distinct sauces, while there is a failure in current literature to recognise the distinctiveness of the blood viscera sauce (Lowe 2017: 309; Botte 2009: 19). The blood viscera garum is recognised as special and high quality but it is rare to see an acknowledgment that it is sufficiently different to have required its own terminology, in production and in trade (as expressed on tituli picti). In the Geoponica we see that liquamen is only used to designate the whole-small-fish sauce and is never seen to indicate the blood viscera sauce in any ancient context, which clearly demonstrates that liquamen does not function as a generic term in relation to this issue. One must always bear in mind that if these terms were used generally in all contexts it would have been impossible to distinguish one product from another. There are other approaches that stress the potential diversity in fish sauce such as the one proposed by Bernal-Casasola (2009), who has identified mixed sauces containing meat, fish and shell fish in amphora residues.Footnote 10 Bernal-Casasola also recognises a ‘garum effect bias’ in current research that is restricting our perception of fish sauce diversity, while I would counter that the potential diversity of fish sauces stems from the creativity of the cook when handling fish sauce before consumption rather than the manufacture of the essential substance (Bernal-Casasola 2016).
In a poem by Manilius which describes scenes of fishing and fish sauce manufacture we can see that the ancients saw these sauces in terms of two basic types. The poem is immensely complex and space restraints prevent a detailed analysis (see Lowe this volume). The catch is initially tuna but these are butchered and cut up and their blood stains the sea (line 666). We then change almost seamlessly to a different scene indicated by tum quoque (then again). This change in scene has not been noted before but is of fundamental importance as I think it also means a change of fish species.
when the catch is brought ashore whole (it therefore cannot be tuna?) and a second slaughter is done to the slain: they are cut into pieces and from the one body different purposes are allotted. One type (i.e. sauce) is better with its juices removed another with them retained.’ (Manilius Astro. 5. 664–72).
Other interpretations have been given,Footnote 11 but I see this as a clear reference to the manufacture of the two types of better quality fish sauce from mackerel (Scomber sp.), one using the juices separately (garum), the other using whole fish including the juices (liquamen). In another change of scene signified by aut cum (676 alternatively when), Manilius describes a sauce derived from crowds of little fish dissolving into what we can assuredly call a more commonplace liquamen (Astro. 5.680).
In the Geoponica garos and liquamen are completely interchangeable terms. Diocletian’s price edict carries the same fundamental information: liquamen is rendered in Greek as garos. The late Roman medical text On Chronic Diseases by Caelius Aurelianus is often quoted to assert that ‘garum is commonly called liquamen’ (2.3.70), but this work is described by Caelius as a direct translation from the Greek of a work by Soranus in the 2nd century AD (Drabkin 1950). The scarcity of the term garum in Caelius’ time (late 4th/5th century) would suggest that he may not have had knowledge of this separate product. He would logically transliterate garos into garum, unaware that they did not always mean the same thing. There is no reference to a Latin garum in any of these three important sources, and the question has always been: if this is all about Roman fish sauce, where is the garum?
The answer is essentially very simple and is concerned with when and how the blood viscera fish sauce was invented and introduced onto the Roman market, and more importantly how long it stayed popular and commercially viable. For there was a time before it arrived and a time when it became less popular, which we need to recognise, if the later scarcity of the term, and in fact the product itself, is to be understood properly.
The remaining evidence for the ‘other’ blood viscera sauce is obscure. It appears on amphora tituli picti alongside liquamen and in elite consumption satire but when and how it came into being is unknown. Pliny the Elder tells a story of an invented garum dated very cautiously to the first part of the 1st century AD. We are told of Apicius’s death occurring prior to AD 42/43 by Seneca (Sen Ad helv. 10.8-9; Costa 1994) but we have no other dating evidence.
Marcus Apicius… thought it especially desirable for mullets to be killed in a garum sociorum – for this thing also has procured a designation (Pliny HN 9.66.4).
This appears to be our earliest textual reference to a sociorum sauce. The excerpt generates a number of questions. Was Apicius responsible for inventing the very idea of a blood viscera sauce or was it already being made and he suggested using mullet and only coined the term sociorum for this fish? Garum was later made from the otherwise unused parts of mackerel and tuna, and tuna would appear to be particularly suitable providing the highest ratio of waste volume per fish. Was the very concept of garum entirely a gourmet led innovation? The impetus for this kind of culinary innovation may have come directly from the elite gourmet community as it clearly becomes very fashionable to use this kind of bloody sauce. How was the idea conceived? Logic would suggest that a manufacturer triggered the process given that it is made from the apparently inedible elements of fish which would otherwise be discarded or buried. Should we consider Apicius’s killing of the mullet just a form of cooking and does that mean garum sociorum was a legitimate cooking medium? These will probably remain unanswered questions.
The use of the obscure term sociorum in this Apicius quote, which is generally rendered as ‘of our allies’ was originally assumed, from a reference in Pliny (HN 31.94), to be linked with trading associations. Leon (2001: 176) has suggested that it is a general term to designate all Spanish fish sauces; but this is another example of the tendency of modern scholarship to see all these terms as generic. It is better interpreted as a very specific name for a very specific sauce, i.e. a mullet blood viscera garum and in this case it is used to cook/kill ‘their fellows’, in other words more mullet. Sociorum is subsequently used by Martial and Seneca to designate a sauce derived not from mullet but from mackerel and its blood (Martial Sat. 2.102; Sen Ad helv. 10.8-9). In this view the Latin term garum was eventually retained and/or appropriated by the elite and the merchants selling these products to designate blood viscera sauces generally; that means one made from the waste products of multiple species as well as specifically mackerel, mullet or tuna. This unusual situation effectively forced the manufacturers of garos to coin a word to designate the primary product in order to separate garum/garos and distinguish it in trade. What also seems apparent is that there was great diversity and specialisation in relation to the species utilised to make these sauces but not, I would argue, over their principle nature.
Once this distinction is recognised it allows us to clarify many puzzling issues but it also raises many questions. The very fact of so many complex contemporaneous garum and liquamen tituli picti on urceii from Pompeii only make sense if they represent distinctive products rather than one being a by-product of the other. However, we must ask how the change in meaning was managed, on the ground, among the traders and merchants. Were there tituli picti designating a garum before the switch in meaning? The dating of tituli picti is not accurate enough to pinpoint the change. These questions must remain unanswered but do require consideration.
A rare bilingual titulus pictus from an amphora from Masada and associated with King Herod the Great has caused many interpretative problems but may shed a little light on this issue. It reads “garum,” two unknown symbols, followed by Bασιλέω(ς). With a putative genitive ending it is said to mean ‘garum of (or for) the king’ but why is it bilingual, and why is it not written as garon or haimation…? I suspect that this label indicated that the content was a blood/viscera sauce not an ordinary garos and the use of the Latin signified this. Herod the Great lived in Rome for some time during the early 40s BC and may have acquired a love of this new sauce. This may explain why this sauce was identified using the new fashionable Latin usage i.e. garum (Berdowski 2008: 116). Herod was dead by about 4-1 BC and so if this interpretation is correct it may be the earliest evidence for the ‘other’ garum in the Late Republic (Cotton et al. 1996).
There is also an increasing body of evidence from papyri for the use of the term garos in Egypt. A flask of garou haimatitou—bloody garos—appears on a fragmentary receipt from the 4th century AD (see note 8). The vast majority of references (57) are to garos alone with occasional references to leukos meaning ‘clear’ or ‘bright’ and in one instance this is listed with another bottle described as simply melanos—black. That garos could be either dark in colour or in contrast paler is apparent from modern experiments and must surely refer to the two types of sauce that were available at the time, i.e. ca. 4th century Egypt, one dark and one light in colour.Footnote 12
In elite Roman consumption texts, garum occurs largely on its own but it is also described as luxuriosa (expensive, Martial Epi. 13.102) as well as arcanum (secret, mysterious, Martial Epi. 7.27.8), yet also nobile (renowned, Martial Epi. 13.82.2). Pliny and Martial also use the term sociorum (Pliny HN 31.93-94; Martial Epi. 13.102). Seneca talks of this ‘so-called garum sociorum, the costly extract of poisonous fish, (which) burns up the stomach with its salted putrefaction?’ and refers back to the cooking/killing of the mullet in the garum sauce (Sen Ep. 95.25; NQ 3.17.2-3). Martial says of his garum sociorum that it is specifically ‘made from the blood of a still breathing mackerel’ (Martial Epi. 13.102) linking the fine blood/viscera sauce with this term at least in the minds of elite consumers.
The perceived absence of the term liquamen in early Latin texts can now be better understood. Fish sauce was originally a singular essential substance: a small-whole dissolved fish sauce and it is perfectly understandable that late Republican and even Augustan writers should transliterate garos into garum as they perceived this product in this way. The blood viscera sauce was invented some time at the end of the Republic but it was much later when this new garum became integrated into culinary use and recorded in culinary, medicinal and veterinary texts where it would be visible to us. The switch in usage was gradual, being consumer-led, and created much confusion which can be seen in the way garum is depicted and utilised in early 1st century consumption texts. We do not see liquamen meaning a specific fish sauce referred to in any kind of text, other than Apicius, until very much later in the empire with Palladius in the 4th century (Palladius Opus Agri. 3.25.12).Footnote 13 The collections of colloquia, daily conversations to aid learning of Greek and Latin, loosely dated from the 2nd to the 5th centuries AD, contain many references to garos which are translated in Latin by liquamen. These texts give information about daily tasks associated with Roman life and we can see how ancient peoples engaged with fish sauce each time they ate. Garum and or an alternative to garos is conspicuous by its absence (Dickey 2012).Footnote 14 These texts reveal that a blood viscera sauce did not figure in daily eating any more.