THE SEARCH FOR A HARBOUR

The triple commission of the fleet, outlined at Agr. 10.4, i.e. the circumnavigation of northern Britain, the conquest of the Orkneys and the voyage to Shetland, should be paralleled at 38.3-4. The circumnavigation is again referred to:

praefecto classis circumvehi Britanniam praecipit

He instructed his fleet commander to sail around (northern) Britain

The landing on the Orkneys is implied, as Furneaux ¹⁵⁵ and Heubner¹⁵⁶ noted, by:

datae ad id vires

forces were allocated for that purpose

But there appears to be no mention of Shetland where there ought to be. The answer lies in the enigmatic and contentious sentence presented thus in the standard editions:

Et simul classis secunda tempestate ac fama Trucculensem portum tenuit, unde proximo Britanniae latere praelecto omnis redierat

This is translated by Handford¹⁵⁷ as:

At about the same time the fleet, which aided by favourable weather had completed a remarkable voyage, reached Trucculensis Portus. It had started the voyage from that harbour, and after coasting along the adjacent shore of Britain had returned intact.

Now any translation is only as good as the text on which it is based. But if the manuscript is corrupt, then so is the text, and consequently the translation is valueless¹⁵⁸ . The above is a well-intentioned, yet seriously flawed, attempt to do justice to Latin which, even translated less freely, would still be virtually meaningless. But introducing into the translation words which are not there in the Latin (had completed a remarkable voyage, it had started the voyage), albeit prosaic licence, is at least better than introducing into the Latin, as some editors have done, extra words (profecta, starting out) which cannot possibly be there. Tacitus would never sacrifice logic to alleged conciseness.

It has often been assumed that the mysterious Trucculensis portus was the winter station of the fleet¹⁵⁹ . But there is no basis for this assumption, apart from the fact that no sensible Roman governor would allow his fleet to spend the winter in some remote spot in the far north of Scotland ¹⁶⁰ . Why would Tacitus bother to mention it, when he doesn't tell us where the army was quartered for the winter ? Rivet/Smith¹⁶¹ were correct in referring to Tacitus' reticence on place names. But to state that Tacitus names it because it was the furthest known place on the mainland will convince no one. The furthest known place was Ptolemy's Tarved(un)um (Dunnet Head) which ultimately derives from Agricola's reconnaissance reports. Besides, how do we know it was on the mainland? Nowhere in the monograph does Tacitus name any military base or even civilian settlement in Britain. Consequently any obscure geographical reference in the Agricola, whether to a harbour or to a tribe, should be regarded with the highest suspicion. Tacitus was not interested in presenting geographical trivia in his biography and we should not be fooled by Dark Age dyslexia.

After the battle of Mons Graupius the combative role of the army was finished. Its installation in winter quarters was not due to the imminent approach of winter¹⁶² , but to the fact that its campaign was virtually over. The battle was won, the enemy was demoralised and hostages were taken from tribes in the extreme north. But the role of the fleet was not over; its triple commission had yet to be implemented. More to the point, the naval arm of Agricola's forces merits the same consideration as his land forces.The fleet cannot compete on the same terms as those in the field; its mainly transport functions relegate it to the status of support. It cannot win a victory over the natives because they have no combat vessels. Its victory has to be over the elements, tempestatum ac fluctuum adversa ¹⁶³ or over Oceanus. It was the threat, posed by the fleet to a people unfamiliar with large-scale seaborne operations, which enhanced its prestige, enabling it to compete on equal terms with the land forces. But the fleet needed an achievement to match Mons Graupius. That achievement would be the conquest of the Ocean barrier to reach Thule¹⁶⁴ . So during the lentum iter¹⁶⁵ of the army the fleet, far from returning immediately to a winter base, was assigned the commission of circumnavigating north-west Scotland from east to west, receiving the submission of the Orkney Islands and reconnoitring Shetland for the reasons I mentioned earlier.

Into this equation comes Trucculensis portus. How do we interpret this curious and contentious location which has spawned a plethora of translations, varying from the "leave it as it is school" of Handford to the buccaneering Port Trucculum of Starr ¹⁶⁶ ; from the Norwood/Watt absurdity¹⁶⁷ , Trucculene Harbour to the equally impossible harbour of Trutulium offered by Church/Brodribb ¹⁶⁸ ? Trucculensis would suggest a noun Trucculum. But there is no such place recorded in any geographical source. We can relate it to a known archaeological site. But unless it occurs in Ptolemy or is attested elsewhere, this would be pure conjecture. We may account for it as a transcription error such as Truxelensis portus or as a circumlocution for a less familiar location recorded elsewhere, such as Tunocelensis portus or Ugrulentum portus or the distortion of a more familiar one, Rutupiensis portus. Since I do not consider any of these to be the winter base of the fleet nor believe that Tacitus would have written anything of the sort, the validity of all four suggestions requires examination.

Truxelensis was implied by Smith¹⁶⁹. Without acknowledging the existence of an alternative MS reading he proposes that some such adjective is formed from a noun Truxulum, itself a cursive (?) script corruption of Fl. Uxelum and says that this river is associated with Uxelum which can be identified with reasonable confidence, through Ptolemy, as the fort at Ward Law ¹⁷⁰ in Dumfriesshire and hence the river name Uxelum must refer to the Nith (he attributes Ptolemy's Novius to the inlet of the Urr Water). He then suggests that the forts at Ward Law and Lantonside functioned as the naval base of portus Trucculensis. Stretching a MS reading to produce a hypothetical noun corruption in order to create a hypothetical adjectival corruption is a rather circuitous route towards establishing a hypothetical naval base which would have been of no interest at all to Tacitus. Smith doesn't explain why Tacitus didn't simply write Uxelum. But why would Tacitus bother to mention such an obscure place?

Tunocelensem was proposed by Hind ¹⁷¹, basing his emendation on Itunocelum of the Ravenna Cosmography. This was aptly disposed of by Rivet/Smith¹⁷² . But it was considered suitable enough to find its way into the variants of the Winterbottom/ Ogilvie edition¹⁷³ . Its dismissal by subsequent editors is understandable. It is hardly likely that an obscure place-name, whose noun form itself is highly dubious, is going to generate an adjective worthy of mention by Tacitus. Hind doesn't explain why Tacitus did not simply write Tunocelum or, better still, Itunocelum (sited in the Solway Firth). His comment that although the best manuscripts (?) read "trucculensem", there is a variant "trutulensem" is curious. The three 15th century MSS derive from the Caroline codex (E), the only one which is the best because it forms the basis of the rest. In fact trutulensem (E2m) has greater validity. To claim that tunocelensem will lead to the clarification of the context and the elimination of a difficult reading is optimistic. I do not see how trucculensem could be a corruption of tunocelensem nor do I find Tacitus' name "trucculum" anywhere in the text. I support the observation of Hanson¹⁷⁴ that this location is highly improbable. The entire sentence, not just one word, requires explanation.

Ugrulentum was suggested by Reed¹⁷⁵ who doesn't provide a precise location except for a vague somewhere north of Loch Broom. Commenting on Richmond's objection ¹⁷⁶ that trutu- has the best authority, he says, this (sc.trutu-) can now be seen from Ogilvie's app.crit. to be untrue; rather, if anything "truccu-" has the best authority. I see nothing in Ogilvie's apparatus to support the superiority of truccu-. In fact Ogilvie¹⁷⁷ points out that the corrector has the same authority as E itself, although even he seems unaware that E is usually inferior to E2m in the spelling of proper names¹⁷⁸ . Murgia suggested that trucculensem was influenced by truculentus (ferocious), an idea taken up by Maxwell¹⁸⁰ . This might well occur to a 9th century copyist, but is hardly a welcoming sight to a sea-sick marine. Till ¹⁸¹ believed that trutulensem was due to the corrector's disapproval of the double -c. Reed's suggestion derives through Furneaux¹⁸² from an idea by Hübner ¹⁸³ who tried to establish a similarity with Ugrulentum of the Ravenna Cosmography ¹⁸⁴ . Unfortunately Reed fails to explain how Ugrulentum could possibly be corrupted to Trucculensem. Dillemann ¹⁸⁵ at least attempted to trace a pattern of distortion, even if it wasn't convincing, and was prepared to admit that any connexion between Ugrulentum and the 'Tacitean' reading was without cogency. But against Ugrulentum is the fact that it doesn't appear in Ptolemy's list. Like the previous two suggestions why would Tacitus bother to mention it?

Rutupiensem (of Richborough) was suggested by Murgia¹⁸⁶ . His route is rutupiensem -> rutulensem -> trutulensem. This is more plausible than the previous three suggestions. The idea began with Rhenanus¹⁸⁷ who suggested Rutupensem. But Rutupiensem, an alternative offered by Lipsius ¹⁸⁸ , should retain the extra vowel from the noun Rutupiae. Lipsius' suggestion found support in the commentaries of Ogilvie/Richmond¹⁸⁹ and Borzsák ¹⁹⁰ . Hind says that this was not likely to have suffered corruption to "Trucculensem" (nor was Tunocelensem), but he doesn't expand on E2m's reading, as Murgia does, and his argument against Rutupiensis on the ground that Rutupinus is the normal adjective is invalid, since his examples are drawn from Latin poetry where Rutupiensis would be metrically impossible. However, the link between such an adjective and a well-known harbour such as Richborough does not explain why Tacitus did not merely write Rutupias, the normal literary usage¹⁹¹ . Hanson¹⁹² produces a convincing case against this suggestion. Furthermore the philological argument is undermined by the logic of naval strategy. With the army wintering in the north it is hardly likely that the fleet would be based in the south-east¹⁹³ , nor that it would begin and end its voyage, as Murgia suggests, at Richborough which it was unlikely to reach until mid-November, when conditions in the Irish Sea and the Channel would be virtually impossible to cope with. If it was planned for Agricola or his successor to renew the campaign the following year (the circumnavigation suggests reconnaissance¹⁹⁴ as much as exploration), then the presence of the fleet in the north is almost certain. We cannot rule out the possibility that the fleet was actively involved off northern Scotland until AD86-87. The taking of hostages in the extreme north and the evidence of permanent military installations, as attested in Ptolemy, as far as the Moray Firth reflects Rome's commitment in the furthest part of the island. Nor can the conquest of the Shetland Islands by Agricola's successor be ruled out. This could be one of the reasons why the Thule episode is understated by Tacitus. Suffice it to say that Agricola potest videri ostendisse posteris ¹⁹⁵.

Some scholars have elected to retain the reading in E, attempting to conjure up ingenious, and sometimes amusing, derivations. Burn¹⁹⁶ offers a Celtic root for truccu-, identifying it with the Welsh Turk or Twrch (boar). Rivet/Smith¹⁹⁷ suggest that the harbour lay at the mouth of a river Truccula or Trucculus, an unsuccessful attempt to dispose of the Ogilvie/Richmond suggestion of Richborough. Then they offer an even more speculative suggestion that the name is based on a Celto-Latin tructa (trout), a river Tructula (little trout), a half-joking name conferred by the Roman fleet. Are we seriously to contemplate zoological derivations which disregard the superior reading of E2m?

Forni¹⁹⁸ , following Anderson¹⁹⁹ and MacDonald ²⁰⁰ , suggested that the harbour was Carpow or Cramond, on the assumption that the fleet sailed all the way around Britain in the middle of winter (as if it had nothing better to do). But the circumnavigation should imply a winter base on the west coast, perhaps at Dumbarton. Hanson ²⁰¹ perceptively saw portus Trucculensis as a point in the extreme north which the fleet reached before returning along the other side of the island to its unnamed base. But why would Tacitus refer to some obscure point in the extreme north of Scotland ? A location on the Scottish mainland is not confirmed by Tacitus. Why does it merit a mention unless its significance were linked to some momentous navigational feat?

So where does all this leave us? Was there ever any likelihood that Tacitus would refer to some insignificant ²⁰² harbour when there were more challenging options for the fleet? Would Tacitus' readers be any the more enlightened by the reference to an obscure location which no one had ever heard of before and was unlikely to hear of again? But Thule is a different matter entirely. The impact of a Roman fleet reaching a harbour in Shetland would be considerable, going some way to exploding a myth, and, as will be shown later, would be reflected in contemporary literature. It is much easier and more logical to assume that the harbour in question is a reference to Shetland, relating to the third leg (the most prestigious in terms of propaganda) of the fleet's commission and reflecting a navigational achievement worthy of record, for whatever reason Tacitus may have underplayed it. I suggest that the main clause should be corrected to:

et simul classis secunda tempestate ac fama trux Tulensem portum tenuit

which may be translated as:

And at the same time the fleet, its ruthlessness enhanced by rumour and favourable weather, reached Shetland harbour.

The adjustment is minimal. Trux (ruthless) in conjunction with the ablative case is relatively common in Tacitus, and if it can be applied to a cohort ²⁰³ and to a battle line²⁰⁴ , it is certainly applicable to a fleet whose effect on the natives would be impressive and intimidating ²⁰⁵. Fama trux relates both to the past record of the fleet and to its anticipated impact on those unfamiliar with its terror, a daunting prospect in Shetland harbour.

The two MS readings merit comparison:

truc-culensem E

tru-tulensem E2m

E is in the habit of confusing c and t²⁰⁶ (common in minuscule). Perret ²⁰⁷ elucidates the problems and makes a strong case for the superiority of E2m's readings of proper names. Till's comment²⁰⁸ on trutulensis that E2m's conjecture has no guarantee is no longer accepted. The corrector, if anything, is guilty of omitting -x rather than of avoiding the double -c. The last three syllables should be identical. Latinised Greek names often drop h²⁰⁹. To create an adjective from an island is easy enough, e.g. Crete -> Cretensis. To describe the harbour as T(h)ulensis portus would be perfectly natural. It did not have a name such as Tunocelum (?) or Ugrulentum, which would only be given when a presence was established there. No other correction comes closer to the now preferred reading of E2m , trutulensem, than tru(x)tulensem. Tacitus' use of simul clearly shows the fleet reaching this harbour at the same time as the army reached its hiberna. A voyage of some fifteen hundred miles²¹⁰ to Richborough would have taken at least two months, if any admiral were rash enough to consider the prospect of such a long voyage in the worst possible conditions. Are we to assume that the lentum iter of the army took this long? If we allow some twelve days for the conquest of the Orkneys and the expedition to Shetland, the army during that time could have covered the distance from Caithness to the isthmus via the Great Glen ²¹¹ where it would encounter the novae gentes ²¹² . Agricola by a display of arms along the western ²¹³ flank of Caledonia might well claim that he had encompassed the whole of it.

The specific mention of a harbour suggests that Agricola's naval officers had seen something to impress them. This would rule out Fair Isle. The most obvious location would be Lerwick of which Nicolson ²¹⁴ says, opposite which lies the island of Bressay, six miles long and curved parallel to the Mainland shore, to provide the natural barrier which makes Lerwick harbour one of the finest in northern Europe. Reaching Shetland harbour is even more impressive than simply reaching Shetland. Cunliffe²¹⁵ suggested that Pytheas may have reached a safe anchorage in Bressay Sound. The Roman fleet had located the same station, which would not be surprising if it had access to Pytheas' marine charts. The fleet's achievement lies in the triumph over nature and the elements, and matches the victory at Mons Graupius. It is noticeable that Agr. 39 begins with hunc rerum cursum (this course of events) rather than hanc victoriam (this victory), a phrase which combines triumphs both on land and on sea²¹⁶.

The introduction of trux eliminates the awkward syllepsis involving secunda. The ablative cases are causal; it was the fine weather and its previous record ²¹⁷ which allowed the fleet to pose such a threat. This was no random phrase by Tacitus; he was looking for something to balance a similar description of the army. The variatio is clear:

fama ferox exercitus²¹⁸

the army, formidable through the report...

The entire clause unde... ...redierat is untranslatable and meaningless as it is presented in all the standard editions, and no one has yet provided a satisfactory explanation for it. If my suggestion of T(h)ulensem portum is correct, then redierat is impossible, since you cannot return from a place you have never visited. The clause must reflect what had happened to the fleet from the time it left Scotland to the time it anchored in Shetland harbour. Since unde with the pluperfect tense always precedes the main verb in Tacitus ²¹⁹, questions must be raised about the validity of unde to see whether it should be retained. An independent clause with the pluperfect tense is normal Tacitean usage and often introduces a reason or an afterthought ²²⁰ . The subordinate clause appears in the MS thus:

unde proximo Brittanniae (Britanniae E2) latere prelecta (lecto E2m) omnis(omni E2) redierat.

The codex clearly poses problems here, producing a mixture of injudicious guesswork and scribal error leading ultimately to bemusement rather than elucidation. But attempts to establish a connection between unde and redierat have created unacceptable linguistic contortions by editors and commentators and must be considered failures ²²¹. Each part of this clause requires close study.

The interpretation of proximo has always produced a dilemma for commentators who couldn't understand what nearest referred to. Tacitus, despite his geographical vagueness, would not have left us to hazard a guess. There is no need to assume, as Burn ²²² proposed, a proximity to the harbour. How can a side of Britain be nearest to a harbour? Ogilvie²²³ refers to the shore adjacent to Rome. How can it possibly be adjacent to Rome ? And where is the word for shore? It is remarkable that in his 'revised' version of Hutton's edition he actually inserted litore (an old conjecture by Pichena) into his text, without bothering to revise Furneaux, whose faulty commentary he used as his starting-point. Ogilvie was trying to establish a connexion between proximo and Richborough, which I have already discounted as the winter base for the fleet in AD83, when the army was based in the north. If the circumnavigation began in the far north east (Caithness, as will be argued later) proximo cannot refer to the southern side, nor to the eastern side, as Gudeman²²⁴ proposed. Tacitus himself had already stated that Agricola's fleet had sailed around the coast of the remotest sea²²⁵ to confirm the insularity of Britain. This must mean the furthest side of Britain, not the nearest. Roy ²²⁶ translated proximo as furthest, trying to reconcile the irreconcilable, while Andresen ²²⁷ made the absurd claim that the nearest side of Britain is the coast of the remotest sea. The furthest side of Britain should be the area from Duncansby Head via Cape Wrath to the Clyde. The possible interpretations of proximo are summarised by Hutton²²⁸ who associates the word with the part of the coast nearest to the fleet; which doesn't tell us anything. Proximo, in my opinion, refers to the side of Britain nearest to the fleet's destination, and if its destination was Orkney and Shetland, then the nearest side is Caithness.

All the modern editions of the Agricola read praelecto. But this is no more than a conjecture, a reconciliation of the two variants, because lecto was considered un-Tacitean ²²⁹. Lecto, if such a form existed, would be all right for Livy, but not for Tacitus, and suggests a conjecture either by E2m or by his exemplar. But praelecta is a different problem altogether because although praelegere is Tacitean usage and, in my opinion, a Tacitean neologism, there is no evidence that in its maritime sense it has a past participle. With the meaning of to coast along, to skirt, to hug, it never occurs in the perfect tense in any prose writer, either in its simple or compound form, for obvious reasons; you reach your destination while coasting, not after coasting. The shoreline doesn't stop after you reach your destination. The perfect tense would imply that the fleet had completed its coasting and was now out in the open sea ²³⁰. Latus praelegens would be acceptable and the present participle of the simple verb is paralleled in Livy ²³¹ . Friis-Jensen, who produced the two past participle parallels from Livy²³² and Propertius ²³³ for TLL, had doubts himself on whether such forms existed ²³⁴. I do not see how you can support one conjecture with two others, particularly when a valid case can be made against all three. Till ²³⁵ said that E's praelecta was the fault of the copyist who had made the participle agree with classis, and he described it as a grammatical impossibility. I suggest that Till was only half correct; the agreement is right, the verb is wrong. Participial corruption occurs in two stages, praeuecta> praeiecta> praelecta>. Praevecta ²³⁶ (sailing on) is an attested form and is more valid than praelecto, a conjecture paralleled only by other conjectures, whose usage is highly dubious and whose introduction creates an ablative absolute where there is no need for one. The fleet had been given instructions to go ahead. Praevecta is here almost equivalent to praemissa, which Tacitus had already used to describe the advance activity of the fleet, praemissa classe quae terrorem faceret ²³⁷.

Ogilvie ²³⁸ says that omnis emphasises that the whole fleet returned intact. But omnis is not used in this sense without a qualifying adjective such as integer or incolumis (without loss). The examples which he cites from Caesar ²³⁹ are qualified by incolumis and cannot be used to support the reading here, apart from the fact that in Caesar it is incolumis, not omnis, which is used predicatively. The predicative use of omnis in the singular is not found in Tacitus nor in any other historian. Incolumis redierat would be perfectly acceptable²⁴⁰ , but not omnis, which here is decidedly odd and uncomfortable. Benario's translation ²⁴¹ , without loss, would imply that Tacitus wrote incolumis. The Rivet/Smith version ²⁴², it had all returned, shows how awkward a direct English translation can be. The corrector (E2) was unimpressed with omnis and removed the terminal consonant to make omni agree with latere (all the nearest side), thus creating, as Ogilvie ²⁴³ says, a weaker sense and an unparalleled word order. This must have been the corrector's own conjecture, not a reading taken from an alternative exemplar. Yet it was good enough to deceive Andresen, Furneaux, Koestermann and Burn. Persson ²⁴⁴ claims that omni is emphatic and that the fleet had sailed round the northern side of Britain, and in fact the whole of it. Apart from failing to offer an explanation for proximo, Persson seems to forget than any completed circumnavigation (which a past participle implies) is hardly likely to involve less than the whole side. Büchner ²⁴⁵ is convinced that the Hersfeld corrector has rightly emended 'omnis' to 'omni', without explaining why 'omnis' ist nicht passende Wort ²⁴⁶. But no good case can be made for creating an isolated omnis and then claiming that it means intact or without loss. Omnis must stand directly next to classis to make any grammatical sense. I don't support Till's argument that its position is emphatic, when he produces no true parallels for a predicative use. If it is retained in its position, it must agree with something else. It might be in the accusative case (= omnes). Its very position in the clause suggests it might be the object of a verb such as adierat with the missing noun being res ²⁴⁷ (resadierat = redierat, the error caused in scriptura continua).

The unde may be partially dispensed with. As may be seen from Till's photocopy of the codex the un- of unde bears a resemblance to -uit of tenuit and would be a dittography. We now have an independent clause. The text will now read,

...Thulensem portum tenuit; de proximo Britanniae latere praevecta omnis res adierat.

...reached Shetland harbour; having sailed on from the nearest side of Britain it had encountered every scenario.

The nearest side of Britain for a voyage to Orkney and Shetland would, as I have already suggested, be Caithness ²⁴⁸ . What Tacitus is saying is that Agricola's fleet had faced both prosperas res and adversas res ²⁴⁹ (ups and downs) from the time it left the coast of mainland Scotland. The prosperas res would be the "discovery" and conquest of the Orkney Islands, the adversas res would be the hazardous sea currents on its voyage to Shetland. But the fleet had faced and triumphed over all the challenges. Omnis res (all eventualities, situations, fortunes) is virtually equivalent to omnem fortunam. The verb adire is normal usage for facing or handling situations, as Livy illustrates with ad omnem adeundam simul fortunam ²⁵⁰ . Res (pl) and fortuna should be interchangeable, both words meaning fortune or circumstances in terms of both fate and finance. The language of the entire sentence is couched in terms as if Tacitus were describing a routine maritime operation. Furneaux ²⁵¹ may well be right in suggesting that the expedition was understated because Agricola had not been personally involved in it. But there is also a possibility that Tacitus was aware that Agricola had merely followed the course plotted by Pytheas, and he couldn't attribute too many initiatives to Agricola in a monograph where his father-in-law had already been overindulged. But in terms of achievement the conquest of the Ocean to reach Thule is equivalent to scaling mount Everest to reach the summit. The objective is in triumphing over the challenge and laying claim to be the first Roman to do so.

Send email to the author: stanwolfson@tiscali.co.uk

INTRODUCTION | PART ONE: Shetland: The Classical Geographical Context
PART TWO: The Manuscripts of Tacitus' Agricola | PART THREE: Reassessment of Vocabulary and Sense
PART FOUR: The Search for a Harbour | PART FIVE: Thule in Contemporary Latin Literature
CONCLUSION | APPENDIX The Boresti: The Creation of a Myth
ACKNOWLEDGEMENTS | BIBLIOGRAPHY | ENDNOTES

Copyright © 2002 by Stan Wolfson. All rights reserved.