1. Causation in the Roman law of delict

Before proceeding to the other requirements of delictual liability under the lex Aquilia, two issues arising from the foregoing discussion still have to be considered. Delict has to do with human behaviour and with a harmful result.

(a) "Factual" and "legal" causation

The first of these two questions can be answered with a straightforward "no". The analysis of delict in terms of objective and subjective requirements, of factual and normative elements and, more generally, of abstract conceptual thinking, is thoroughly un-Roman. The issue of causation provides a good example. That a defendant cannot be held liable unless his conduct has in fact "caused" the injury is one of the few more or less universally accepted propositions of the modern law of delict. But what does that imply? When can the relevant causal connection between conduct and injury be said to be established? Hosts of theories have been developed to address the "never ending and insoluble problems" arising from these questions; theories which have, in view of their "ugly and barely intelligible jargon",²⁴⁹ not seldom even contributed to the "uncertainties and confusions which continue to surround the legal use of causal language".^[5140] Thus, one often encounters a very basic distinction between factual causation and legal causation.^{[5141] [5142] Factual causation is usually defined in terms of the sine qua non ("but for") test.252 To apply it, one has to ask whether, but for the defendant's (wrongful) conduct, the harmful result would have occurred.}

This is where a second unquestioned premise of the modern law of delict comes in: the defendant cannot be liable merely because his conduct has "in fact" caused the harmful result. Some means must be found of limiting his liability.^[5146] Fault is not sufficient for this purpose, for it would hardly be acceptable either, from the point of view of social policy, to establish liability wherever harm has been culpably caused. Such a regime would stifle, for instance, all forms of lawful competition. What is required, is "a policy-based mechanism for eliminating from the causal net these factual consequences for which it would be unreasonable or undesirable to impose liability".^[5147] Should one^[5148] therefore limit the actor's liability to those consequences that were reasonably foreseeable at the time of acting?²⁵⁹ Or perhaps rather to those that flowed directly from his conduct, without the operation of a novus actus interveniens (irrespective of whether the specific result in question was reasonably foreseeable)?²⁶⁰ Or should one require the causal bond between conduct and harmful result to be "adequate" in the sense that, according to common experience, the result can be regarded as a normal or natural consequence of the conduct?²⁶¹ Does one have to focus on the purpose of the rule violated and eliminate all those consequences that are not covered by its protective scope?²⁶² Or would it be more appropriate in this context to activate the requirements of wrongfulness and fault and to ask not (as has traditionally been done) whether the defendant's conduct was wrongful and culpable in abstracto, but whether it was wrongful and culpable in relation to the harm complained of?²⁶³ Does one have to establish (and limit) two causal connections: namely that between the defendant's conduct and the harmful result (existence of liability) as opposed to that between the harmful result (for instance: the injury to bodily integrity or life) and the resulting damages (extent of liability)?²⁶⁴ These are a few of the criteria suggested and some of the issues canvassed in modern legal literature.

The leading case embracing the foreseeability test is The Wagon Mound (supra, note 249). For a detailed analysis (and a proposed restatement) of the foreseeability test, seeJ.C. Smith, Liability in Negligence (1984), pp. 91 sqq., 131 sqq.

Leading case (dethroned by The Wagon Mound) In re Polemis (In re an arbitration between Polemis and Furness, Withy & Co. Ltd.), [1921] 3 KB 560 (CA), but cf. Bacon's maxim "in jure non remota causa sed proxima spectatur" which is analysed byjoscph H. Beale, "The Proximate Consequences of an Act", (1919-20) 33 Harvard LR 633 sqq.

²³ On the so-called adequate cause theory cf, for example, Lange, Schadensersatz, pp- 57 sqq.; Hart and Honore, op. cit., note 250, pp. 465 sqq.

Prevailing view in modern German law; cf, in particular, Ernst von Caemmerer, "Das Problem des Kausalzusammenhangs im Privatrecht", in: Gesammelte Schriften, vol. 1 (1968), pp. 395 sqq.; Lange, Schadensersatz, pp- 76 sqq.

²³ This approach is strongly urged by Boberg, Delict, pp. 274 sqq., 380 sqq. On the difference between the abstract and the relative concept of negligence, cf. the clear analysis by W.H.B. Dean, "Culpability or Remoteness", (1974) 91 SALJ 47 sqq.

²³¹ "Haftungsbegrundende Kausalität" as opposed to "haftungsausfiillende Kausalität"; cf. e.g. H;.rt and Honore, op. cit., note 250, p. 85; Lange, Schadensersatz, pp- 54 sq.; Zimmermann, 1980 Jurist enzeiturtq 16.

^2f’⁵ Cf. eg. Andrews J in Pahgraf v. Long Island Railroad Co. (1928) 248 NY 339 at 352, 162 NE 99 at 103: ".,. because of convenience, of public policy, or a rough sense of justice, the law arbitrarily declines to trace a series of events beyond a certain point. This is not logic. It is practical politics."

²⁶ This point is made particularly clearly with regard to Roman law by MacCormack, Studi Sanfilippo, vol. I, p. 283.

from the Roman sources would therefore be a doubly inappropriate endeavour.

(b) The Roman approach

Of course, the Roman lawyers did not grant an action, if there was no "causal" connection between harmful result and the defendant's conduct. But this was more a matter of common sense than of conceptualizing and applying logical or philosophic notions.^2(S7 There is no evidence for the use by the jurists of a specific theory of causation. Even the phrase "causam mortis praestare" was used as a term of common parlance rather than as a vehicle for the reception of Greek philosophical discussions on causation;^{[5149] [5150] and the idea of a conditio sine qua non, though known and used as an argument to establish or to reject liability in other contexts,26y hardly ever surfaces in our texts dealing with the lex Aquilia. These texts always take for granted that the defendant's conduct did cause the injury. What remained to be discussed was merely whether the specific manner in which the injury had been brought about justified the application of the lex Aquilia or merely of an actio in factum.[5151] The answer to this enquiry, in turn, did not depend on considerations of causation either. The narrow scope of chapters one and three was not attributable to the fact that the Romans of the 3rd century b.c. should have been unable to conceive of the idea of indirect causality;[5152] [5153] it was merely determined by the (linguistic) meaning of the operative verbs.2 2 "Occidere, mere, frangere, rumpere" all implied, as a matter of course, that death or injury had to have been "caused" by the defendant, but so did "causam mortis praestare".}

(c) Concurrent causation

What did this approach imply with regard to notorious problem areas such as concurrent causation?^{2 4} Take the case of a slave who is mortally wounded by A. Before he dies, however, he receives another wound from B which immediately kills him. Application of the conditio sine qua non test poses a problem. If we put the test question: "Would the slave have died but for A's conduct?", the answer is clearly "Yes". The same applies if we ask: "Would the death have occurred if B had not intervened?" Neither A nor B, therefore, seems to have caused the death: hardly an acceptable result. It could be avoided by accepting a sine qua non relation, if the harmful result would not have occurred in this specific manner and at this specific moment without the interference of A (or B ). But would one then not have to hold a surgeon liable who has "caused" his patient's death in the course of an unskilful operation, even though the patient would not have survived a skilful operation either?^{[5154] [5155] [5156]}

The Roman lawyers were not hemmed in by this or any other dogma and decided the case according to their understanding of the word "occidere". "Celsus scribit", reports Ulpian,^[5157]

"si alius mortifero vulncrc percusserit, alius postea exanimavcrit, priorem quidcm non teneri quasi occiderit, sed quasi vuineravcrit, quia ex alio vulnere periit, posteriorem teneri, quia occidit.

Undoubtedly, according to Celsus, Marcellus and Ulpian, B has committed "occidere". A, on the other hand, has not directly brought about the slave's death. B's action intervened and (as a modern lawyer might say)^[5158] "broke" "the chain of causation". A's conduct can therefore not be labelled "occidere"; liability can arise only under chapter three of the lex Aquilia. But this solution was not undisputed. Julian took the view that both B and A were liable under chapter one: "rursus Aquilia lege teneri existimati sunt non solum qui ita vulnerasscnt, ut confcstim vita privarent, sed etiam hi, quorum ex vulnere certum esset aliquem vita excessurum."²⁷¹¹

It is the action of inflicting a fatal wound as such that must be termed "occidere"; as long as death is certain to result from it, it does not matter whether it occurs immediately or only after a while.^{[5159] [5160] [5161] [5162] [5163] [5164] [5165] This conception of occidere, incidentally, was also the reason why Julian calculated the period determined in the "quanti id in eo anno plurimi merit" clause from the moment the slave received the mortal injury.[5166] At this stage he had already committed occidere. Celsus, in line with the argument advanced in D. 9, 2, 11, 3, regarded the moment of death as crucial.[5167]}

2.