ANCIENT SYSTEMS OF DREAM-CLASSIFICATION

BY

A. H. M. KESSELS

I

One of man's least explainable experiences is his dream-life.

This is certainly the case with people of a modest cultural back-

ground, for whom the rational explanation of things has only just started. This quest for a rational explanation did not start in the

same age for all known phenomena. It was only in classical antiquity that serious efforts were first made to provide a rationale for most

of them. This is certainly true of the problem of the dream. It should be borne in mind that it is a great achievement when

man begins to realize that between the dream-world and the

waking-world exists a very fundamental difference. What this

difference really is, and how it can be established, is of course

another question, that is beyond the scope of this investigation. It may only be remarked in passing that even in antiquity one

meets awareness of the problem that strictly speaking there is no

real criterion to distinguish between these two worlds 1). Of course from the above it cannot be concluded that 'originally'

there was a situation in which primitive man did not mark any difference whatsoever between dream-life and waking-life. Still

it is true that the same value was assigned to the things experienced in dreams and the things that happened in waking-life. In any case man regarded the dream as a real event, which was in fact as

'true' as ordinary events 2), even to such an extent that people

1) Cf. Plato Theaet. 158 b-c. For a modern treatment of the problem from a philosophical point of view cf. N. Malcolm, Dreaming (London 1962).

2) Cf. J. Hundt, Der Traumglaube bei Homer (Greifswald 1935), passim; H. J. Rose, Primitive Culture in Greece (London 1925), 152 ; L. Lévy-Bruhl, L'expérience mystique et les symboles chez les primitifs (Paris 1938), 98 ff., who gives a general treatment of the primitive attitude to dreams; cf. also La mentalité primitive (Paris 1922), 94 ff. by the same author. L. Binsw anger

390

were able to show some visible proofs for the reality of the dream-

experience. So at least Pindar in his 13th Olympian Ode (63-78) wants us to understand the tale of Bellerophon, who in a dream

was presented with a 'horse's charm'. After his dream he actually found on awaking the bit lying near him as a visible proof of the dream's reality i) .

Quite early in history man started trying to interpret his dreams.

He did not ask himself in the first place how a dream came about, for this was and remained unexplainable and was therefore ascribed

to the influence of the gods. The question asked was="What can I

do with it ?". The interpretation in practice turned out to be an

effort to find a clue as to the possible relation of the dream with the

future fate of the person concerned. This appears to have happened

already in Babylonian times. People then did not bother much about the how and why of dreams; they were satisfied with an

interpretation which had some practical value.

In Greece, too, for quite a long time dreams were looked upon as a divine sign concerning man's future, notwithstanding several efforts at rational explanation by e.g. Heraclitus, Democritus,

Plato, Aristotle, and Hippocrates. One has only to remember

how the incubation sanctuaries flourished. In Greece, therefore, there existed two methods of approach

to the phenomenon of the dream 2). One was a philosophico-

Traum und Existenz (= Ausgew. Vovtydge und Aufsdtze, I, 74-97) speaks about "die urspriingliche enge Zusammengeh6rigkeit von Gcffhl und Bild" (p. 82) and refers to the dream of Penelope about the geese. On Aeschylus' Persae igi-i96 he remarks (p. 87) : "Man sieht diesem Bild als solchem nicht an, ob es einem Traum oder einem Geschehen in der dusserenwelt entstammt; so verwischt sind bei den Griechen die Grenzen zwischen dem innern Er- lebnisraum, dem dusseren Geschehensraum und dem kultischen Raum".

1) For other instances cf. Paus. 10, 38, 13 and the inscriptions from the incubation sanctuaries, Verg. Aen. 8, 42 and 81 ff. (Aeneas finds a white sow as a proof of the validity of the dream). Prof. Waszink reminded me of the proem of Hesiod's Theogony (22 ff.), where Hesiod relates how he recieved his from the Muses. However, it must remain uncertain whether he actually met the Muses in a dream. There is no trace of this in the Theogony itself, but later writers, e.g. Callimachus, followed Hesiod's story and clearly made it into a dream.

2) Accordingly D. del Corno, Ricerche sull' onirocritica greca, Rend. Ist. Lombard. 96 (1962), 334-366, draws a distinction between 'onirologia' and 'oniromantica'.

391

psychological approach, the other a more practical one 1). These

two remained existing side by side all through antiquity, the first

one chiefly among the more enlightened circles, the other one

being effective in broader layers of the population, even with some

influence on the so-called enlightened minds.

Dreams were divided into different categories and classes. I shall

review these divisions and classifications and try to establish

what connections can or cannot be traced between them.

II

Let us start with the only professional dream-interpreter from

antiquity whose work on dreams has come down to us as a whole,

namely Artemidorus Daldianus 2). It is commonly stated that he was a philosopher of Stoic origin 3).

I must, however, raise objections to this description. In fact only the

Suda calls him a philosopher, and not even a Stoic one at that.

Artemidorus was merely a very practical man, an actual dream-

interpreter, who wrote down the cases of dream-interpretation in which he was involved every day. It would go too far to call such

a man a philosopher. Scholars seem to have been misled by the fact that there are

many Stoic elements in the theoretical parts of his work. But this is only natural, as the Stoics were greatly attracted by divina- tion in general and dream-interpretation in particular. However, the theoretical parts are far outweighed by sometimes rather

boring enumerations of all sorts of dreams. This indicates that

those theories did not really have Artemidorus' interest. In the

1) Under this practical approach I count also the medical application and interpretation of dreams, if only for the sake of convenience. Of course there is a difference between the way of interpretation of e.g. Artemidorus and the medical one, but the latter, too, had a predominantly practical character.

2) The man really was born in Ephesus. But he wanted to use this cognomen to do honour to his mother's native city, as Ephesus was already famous enough because of so many of her citizens. This cog- nomen may also distinguish him from his namesake, the geographical writer, who also came from Ephesus.

3) Cf. O. Stählin-W. Schmid, Geschichte der griechischen Literatur (Mün- chen 1924), II 2, 805, note 1 ; A. Lesky, Geschichte der griechischen Literatur (Bern 1963), 898 ; W. Sontheimer, Kleine Pauly (Stuttgart 1964), I, 617-8.

392

course of this article I hope to show another, and to my mind

rather compelling, reason why Artemidorus cannot be regarded as a Stoic philosopher.

The first remarkable aspect of his work is the fact that Arte-

midorus draws a sharp distinction between the non-

predictive dream, and 'G'veLpoq, the dream that has a particular

meaning for one's future 1). For him of course only the 6veipoi are

of value. He divides them into which find their

fulfilment in just the way they are seen, and which

represent the future in a more obscure way. His whole work there- fore deals almost exclusively with these two types, of which the

are naturally more often discussed. So it is not very

surprising that he mentions the other types only incidentally. These observations refute, I think, the view proposed by Behr2),

that Artemidorus adopted the division made by Plato in his

Republic 3). According to Behr, Plato had 'a two-fold system', for

he knew both predictive and non-predictive dreams 4). Now the division of dreams into these two categories (as far as

Greece is concerned) is at least as old as Homer, who gives a

mythical explanation of them in the Odyssey 5). Furthermore Plato

was not primarily concerned about giving a systematic treatment of

the essence and origin of the dream. Besides, what we know about

Plato's conceptions of this phenomenon, derives only from some

scattered remarks we find throughout his works, of which the

passage in the Republic certainly is the most important one. It was

Aristotle who was the first to give a systematic treatment of this

subject from a philosophical point of view. In Republic 571 Plato sets out to describe the tyrant. The problem

1) Although Artemidorus seems to have forgotten this distinction in two places: 236, 9 and 267, 3 Pack. This may indicate that such a distinction was not made in spoken language, and also that Artemidorus here has fallen back on this current use of language. He himself in fact explicitly says that

and were the technical denominations (cf. 238, 20-239, 14 Pack).

2) C. A. Behr, Aelius Aristides and the Sacred Tales (Amsterdam 1968), 174, note 10.

3) 571 c-572 b. 4) Behr, op. cit., 173. 5) Od. 19, 560 ff.

393

is how such a man can develop out of the capacities which are

part of every human being, and whether he lives happily or not. To

this effect Plato has to examine the nature of man's desires. In

every human being there are terrible, wild and unnatural desires

that express themselves only in dreams (£v 'U'nvoLq

1). If we leave alone the term 'classification' used by Behr, we should

expect from Plato not a two-fold but a three-fold system, with a

glance at his tripartition of the soul. And in fact Plato seems to

know three kinds of dreams, originated respectively by each of the

three parts of the soul (Rep. 571 c-572 b). Of course, only the

dreams that were due to the working of the xoyia<ix6v were relevant

to Plato. It might be argued that, as far as relevance is concerned, there were indeed two categories of dreams. But this is not the

criterion from which we are to judge a dream-classification, for in

that case practically all classifications would turn out to be based

on only two categories of dreams, viz. relevant and non-relevant

dreams. Moreover, if Plato really developed a strict system, he

would not have used the terms 6veipoq and È;vÚ7tVWV (beside some

others) so undiscriminatingly. In addition there are some points of

difference between Plato's remarks on dreams in the Republic

(571 f.) and those in the Timaeus (71 d ff.). It is safer, therefore, not

to credit Plato with any coherent system of dream-classification.

Another, and to my mind even more compelling, reason to keep Plato's remarks and Artemidorus' system separated (a reason

which Behr fails to see) will appear in the course of this article.

Now Artemidorus has a division of dreams into two categories which are again subdivided into 5 classes in all. As has already been said, he knows the category of 6veipoi, which he subdivides

into 'O'vetpoL, and and the category of

subdivided into and 2). As even Behr admits, this five-fold division was a system actually

in existence at the time of Artemidorus. However, this classifi-

1) Rep. 572 b. In Plato's remarks one may find a precursor of Freud's theory of the censor ; cf. 1. Dambska, Le problème des songes dans la philo- sophie des anciens Grecs, Rev. Philos. 15I (1961), II-24, espec. 14.

2) Artem. 179, 13-18 Pack.

394

cation has not been created by Artemidorus himself, but originates from a much earlier time. Indeed the had already been the subject of treatises written by Artemon from Miletus, De- metrius of Phalerum and Geminus of Tyre. For doubtless we may equate the crUV1"(xY(X[ and 6?P«?Ei«a sent by Sarapis 1) with the

so-called as well as the 6EÛ>V of which Arte- midorus gives a treatment in the 22nd chapter of his fourth book. Besides Artemidorus notices that this same Artemon as well as

Phoebus from Antioch had written about the 2). Moreover we know that 6pmym was used in connection with dreams

as early as the third century B. C. 3). Perhaps itself

may not be found in early papyri, but certainly is used

of dreams giving indications about the future, both in papyri and other early texts 4).

1) Artem. 323, 21 Pack. Cf. Wilcken, U.P.Z. 1 20, 27 ; 1 18, 30; D. del Corno, Graecorum de re onirocritica scriptorum reliquiae (Milan 1969), 110.

2) Artem. 6, 14-16 Pack. Cf. D. del Corno, op. cit. 112, note 7, who suggests that Artemon may have been the author of this classification (cf. ibidem, 175, note 27). This hypothesis must remain uncertain, especially as the terms and were used long before Arte- mon's days (ist cent. A.D.) in connection with dreams; cf. A. Wikenhauser, Die Traumgesichie des Neuen Testaments, Antike u. Christentum, Ergän- zungsband I (Münster 1939), 332-333; for cf. Arist. E.N. 1102 b 10 ; P.N. 458 b 18 and 463 a 29.

3) Cf. Pap. Goodsp. Cairo 3, 5. In Wilcken's U.P.Z. 1 78, 37 we find it used in the second century B. C. Besides there are quite a few places in the Septuagint where is attested in this sense.

4) I.G. XII 2, 108 however, is of an uncertain date. in connection with incubation appears in the third century B.C., cf. Pap. Cair. Zen. 59, 034 (dated 257 B.C.) lines 4-5:

(cf. P. M. Fraser, Two Studies on the Cult of Sarapis, Opuscula Atheniensia III (Lund 1960), 1-54); I.G. XI 4, 1299 (about 200 B.C.), line 13:

Pap. Par. 3209, end: This seems to corroborate Th. Lefort's theory about

the practice of incubation, especially in the sanctuaries of Asclepius, which he has developed as a supplement to Deubner's work (cf. Le Musée Belge 10, 1906, 21-38 and 101-126). He considers the oldest form of incubation to be the practice of sleeping on a holy ground (by the sick) in the belief of re- cieving cure during sleep. This was modified later on (after the 4th cent. B.C.) into the practice of sleeping in temples with the end of recieving methods of healing or other commands from the gods: then would be very appropriate. For other instances of the verb cf. Mt. 2, 12 and 22; Aristid. 50, 5 Keil; Jos. Ant. XI 327.

395

Two centuries after Artemidorus we come across the same

classification in the work of Macrobius 1). He gives the following translations of the Greek terms: ov?cPo? = somnium, 6pmym = visio,

= oyaculum, £v4<viov = insomnium, and =

visum. At this last equation he notes that this in fact is Cicero's

translation 2). The classifications of Artemidorus and Macrobius

show such a degree of agreement 3) that it is quite certain that

they both directly or indirectly made use of the same source.

As we have already noticed, Artemidorus divided the ov?cPoc into

and «aa?yopcxoi. Of this distinction no trace can be

found in Macrobius. But it really is remarkable that Macrobius'

definition of visio (visio est autem curn id quis videt, quod eodem modo

quo apparuerat, eveniet) bears much resemblance to Artemidorus'

conceptions about the 6veipoq However, a de-

finition of 6pmym = visio Artemidorus has not given. No less did

he define what exactly should be understood by But

Macrobius does give a definition : et est oraculum quidem cum in

somnis parens vel aliqua sancta gyavisve persona seu sacerdos vel

etiam deus aperte eventurum quid aut non eventurum, faciendum vitandumve denuntiat 5). In Artemidorus we find a parallel in his

remarks on the persons that appear in dreams and whose messages

ought to deserve some credence 6 ) . So we can percieve some points of difference in their classifi-

cation-systems. Now it has to be noticed that the division of

6veipoi into «aa?Yopcxo? and is not attested anywhere else. Moreover it is only in Artemidorus' dream-book that

is used in the sense of 'to be interpreted as seen' 7). All this

1) Comm. in Somn. Scip. 1 3, 2 ff. 2) 1 3, 2 quod Cicero, quotiens opus hoc nomine fuit, visum

vocavit. Cf. Cic. Acad. I 40; II 18 ; II 88, and J. S. Reid's commentary, pp. 233 and 283.

3) For further points of agreement between Macrobius and Artemidorus about the interpretation of dreams cf. C. Blum, Studies in the Dream-Book of Artemidorus (Uppsala 1936), 53-56.

4) Macr. Comm. I 3, 9, cf. Artem. 4, 23 and 241, 2-3 Pack. 5) Macr. Comm. 13, 8. 6) Artem. 195, 3-196, 18 Pack. 7) However, the Suda s.v. mentions this division, and it is

adopted by Eustathius ad Hom. Od. 19, 558 in his interpretation of Pene-

396

makes it acceptable that this was Artemidorus' own invention.

Because his class of 6vetpoL apparently coincided with the usual meaning of 6pmym, this distinction has not been main- tained after him. With he did almost the same thing. Having no use for its definition in his work he disconnected it from

this sort of dreams and fitted it into his work again at a different

place 1). For it was rather obvious that dreams as defined by did in fact happen in Artemidorus' practice as a

dream-interpreter, and were in need of interpretation. For indeed

such appearances of gods and other important persons in dreams will doubtless not have been restricted to such dreams as occurred in incubation sanctuaries. They certainly formed part of the general cultural pattern of those times 2).

'

III

As might be expected, the above is not the only classification-

system of dreams that existed in antiquity. That is why we now

have to take a closer look at the system of Cicero, who adopted it

from Posidonius. Cicero mentions this system in his treatise about

divination: sed tribus modis censet (sc. Posidonius) deoyum adpulsu homines somniaye : uno quod ?yovideat animus ipse per sese, quippe

qui deorum cognatione teneatuy, alteyo quod plenus aey sit inmoy-

talium in quibus tamquam insignitae notae veyitatis

appareant, teytio quod ipsi di cum doymientibus conloquantuy 3). Posidonius' system has equally been adopted by Philo Judaeus, of

course with some adaptations, because Philo wanted to explain the

dreams he found in the Old Testament by means of this system. This classification is to be found in his treatise Tcep'L -rou 6zo7tz!L'!'ouç elvmi TouS ovdpouç 4).

lope's dream. Both writers derive their knowledge of this division from Ar- temidorus (at least, I cannot think of anyone else), which shows that Ar- temidorus had a good deal of authority even in the tenth or twelfth century.

1) Cf. the preceding note. 2) Cf. E. R. Dodds, The Greeks and the Irrational (Berkeley 1951), chapt. 4. 3) Div. 1 64. 4) It is possible that Posidonius ascribed a godlike character to all pre-

dictive dreams, but not to all dreams indiscriminately. Cicero (Div. 1 64) not without good reason uses the words deorum adpulsu, just as Philo writes

397

Cicero's first category corresponds with Philo's third one:

Cicero's second category we find again in Philo in a somewhat altered form: 8zú'!'zpov &' d80ç, Èv 0 Ó vouç '!'wv lxmv

elvmi xml 7rPoywt?6xEw Tt !LZÀÀÓV'!'ú)V 2), and

also To 8È: 8zú'!'zpov Tcw oacw <TUYXt.?OU?.SV?<; XOH 6EO<pOp7jTOU 7tOÀÀLl

<pomyope£ew 3). That this in fact does go back to the same origin as Cicero's second category has already been shown

by Blum 4). Cicero's third category Philo defines as follows: Twv

Time and again efforts have been made to reconcile the five-fold

system of Artemidorus/Macrobius with Posidonius' three-fold

one'). It has to be admitted that the five-fold system also com-

prises the two classes of non-predictive dreams, and gav- ,rcxa[tv,7(xl and if we leave these out, a three-fold system is what re-

mains. But can it be proved that these two three-fold systems have the same origin ? a

1) II 1. 2) I 2. 3) II 2. 4) Op. cit., 66 ff. His further conclusions, however, are objectionable and

in my opinion inexact, as I hope to demonstrate in the course of my argu- ment.

5) I 1. 6) II 2. 7) A notable exception seems to be L. Deubner's De incubatione capita

quattuor (Leipzig 1900). In this work the various relations between the systems of dream-classification found in Artemidorus, Macrobius, Nice- phorus Gregoras, John of Salisbury, and the pseudo-Augustinian work De Spiritu et Anima are discussed. Deubner does not mention Posidonius' system anywhere in connection with Artemidorus' classification.

398

Blum 1) thought that there was indeed a strong resemblance

between the two systems. He started from Philo II 3, where we are

told that dreams directly sent by the gods can be compared with He then supposes that these are wholly identical

with the oracula of Macrobius, and to this end quotes Macrobius'

definition cum in somnis payens .... vel deus aperte eventurum quid aut non eventurum.... denuntiat 2), putting much emphasis on the

word aperte. For xpqayol ampeiq, however, one could as well take

into consideration the theorematical dreams of Artemidorus or

Macrobius' visio. Philo, incidentally, speaks about not

about the verbal messages that Macrobius seems to have meant by denuntiat. Therefore 6«cpsCS only indicates that such dreams

are clear at once 3). Dreams that are caused by the soul's own power, i.e. Cicero's

first category, and which need the art of a dream-interpreter, are

put by Blum on an equal footing with the somnia of Macrobius 4). This does not necessarily have to imply an agreement of both sys- tems, for it was generally assumed that not all dreams were clear

at once, and that some of them had to be interpreted by such men

as Artemidorus.

With the second category, however, Blum is somewhat at a

loss. He notices that it is not quite identical with Macrobius'

visio. Then he thinks that he has found a way out by remarking that a specific instance of visio in Macrobius' conception cannot

be found in Genesis, which forms the basis of Philo's treatise on

dreams, and that this is the reason why Philo altered it somewhat 5).

1) Op. cit., 67-71. 2) Comm. I 3, 8. 3) M. Gelzer, Zwei Einteilungsprinzipien der antiken Traumdeutung, in

Iuvenes dum sumus, Aufsätze zur klassischen Altertumswissenschaft der 49. Versammlung deutscher Philologen und Schulmänner (Basel 1907), 49, gives an interpretation which is in fact the same as the one put forward by Blum: "Nach beiden wird der Träumende durch Worte über seine Zukunft auf- geklärt". Gelzer further identifies Philo's third class with Macrobius' som- nium, because both need interpretation, and Philo's second class with visio. Certain points in Gelzer's interpretation have already been refuted by J. H. Waszink in Mnemosyne III g (1941), 81 note 3.

4) Comm. 1 3, 10. 5) Op. cit., 68.

399

This argument, however, is not consistent if one considers the

fact that Cicero made use of the same system as Philo. Then with

him, too, the second category should correspond to Macrobius'

visio. Of this Cicero does not give any indications, although of

course for his work on divination he could have made use of a

much more extensive literature than Philo, and therefore may have

found a typical example of visio very easily.

IV

It is now, I think, sufficiently clear that these two systems did

not originate from the same source 1). Besides there is another and

even more compelling reason why these two systems are irrecon-

cilable.

The point is that all classifications are answers to certain

questions. In this case the questions from which they started are

totally different.

The classification of Artemidorus/Macrobius is an answer to the

following problem: one sees certain things happen in a dream; are they signs of future events or are they not ? And if they are, how are those signs expressed; if they are allegorical signs, we call

the dream 6veipoq ; if the future is told by means of some important

person appearing in the dream, we call it and if we

foresee the future in our dreams in exactly the same form in which

it will occur, we call it 6pmym. If the signs in our dreams, however, are not signalling future events, those dreams may be either

or For Macrobius again gives an

explanation in answer to the question: "What do I see in the dream

and what does it mean ?" 2). For however, Macrobius 3) gives an explanation as to

how these dreams originate. This shows that the common expla- nation of non-predictive dreams was accepted by the dream-

1) Another objection to the supposed interrelation between Posidonius' and Artemidorus' systems is based on the terminology in which the various classes are described by Artemidorus. For where did he find this terminology ? Neither in Cicero's works nor in those of Philo is there any trace to be found of this terminology.

2) Comm. I 3, 7. 3) Comm. I 3, 4-6.

400

interpreters, too. It was an old explanation, traces of which are to be found in Herodotus (VII 16), Plato and the various medical schools.

The classification of Posidonius is totally different, because it

tries to give an answer to the question: "How is it possible that human beings (with the aid of God) are able to get a certain know-

ledge of the future in their dreams As we have seen, this ques- tion recieves three answers.

However, it is not quite clear whether Posidonius accepted all dreams as signs of the future. Perhaps the phrase deoruns

adpulsu in Cicero indicates that Posidonius did not exclude a

category 1) of dreams which were not under the influence of a deorum

adpulsus, and which therefore did not have any predictive value 2). This distinction leads us back to what we have noticed at the

beginning of this article: in ancient Greece there were two ways of approaching the phenomenon of the dream, namely a practical one, manifesting itself in the classification of Artemidorus/Macro- bius, and a philosophico-psychological one of which Posidonius is

an exponent. This philosophical classification also appears in a christianized

form in Tertullian. It is clear from his definitions that his classifi-

cation does not refer to prophetic dreams only. It goes as follows:

definimus enim a daemoniis Plurimum incuti somnia, although these

are vana et fyustyatoyia et turbida et ludibriosa et immunda... A deo

autem... ea deputabuntur quae ipsi gyatiae compayabuntur, si qua honesta sancta pyoplaetica revelatoria aedificatoyia vocatoyia... Teytia

species erunt somnia quae sibimet ipsa anima videtuy inducere ex

intentione circumstantiarum 3). Waszink 4) has shown that the source of Tertullian must have

been Hermippus of Berytus, of whom nothing much is known

except that he lived under Hadrian and wrote about dreams.

It is very likely that Tertullian kept very close to the text of

Hermippus. The conclusion must be that the original idea of

1) Div. I 64. 2) Cf. above, p. 396, note 4. 3) De A nima 47, 1-3. 4) In his commentary on De Anima (Amsterdam 1947), 45*.

401

Posidonius of classifying only predictive dreams was gradually lost 1).

V

Another classification has given rise to even more problems and

therefore to even m.ore solutions. I mean the classification which

Calcidius has drawn up in his commentary on the Timaeus 2). It

seems best first to cite it in full: Consentit huic Platonico dogmati Hebyaica philosoPhia; appellant qui?pe illi vayie, ut somniu-m et

item visum, tum admonitionem, etiam spectaculum nihiloque minus

revelationem: somnium quidem, quod ex reliquiis commotionum animae diximus oboyiyi, visum veyo, quod ex divina viytute legatur, admonitionem, curn angelicae bonitatis consiliis regimur atque ad-

monemuy, spectaculum, ut cum vigilantibus offeyt se videndam caelestis

potestas claye iubens aliquid aut prohibens forma et voce miyabili,

revelationem, quotiens ignoyantibus sortem futuram imminentis

exitus secreta panduntur. When he tells us that the Hebyaica philosophia is in accordance

with the teachings of Plato, as far as dream-interpretation is

concerned, Philo is meant, as is commonly accepted 3). Alas, this

appears to be the only thing that can be said with certainty about

his classification. The efforts that have been made to reconcile the

classifications of Calcidius and Macrobius have resulted in very

divergent and to my mind unsatisfying solutions. Let us survey them briefly.

Blum 4) proposes the following identification: admonitio -

spectaculum = revelatio = 6veipoq ; somniurn =

Evunwov ; but "the description of visum does not tally with that of / 1)

1) For the other Christian adaptations of this system cf. Waszink on Tertullian's De Anima, p. 501 f. After my manuscript had been finished D. del Corno's work on the fragments of Greek onirocritical writers was pu- blished (cf. above, p. 394, note 1). He has come to similar conclusions about the (non-existing) interrelation between Posidonius' and Artemidorus' classifications (cf. p. 173 ff.).

2) Comm. in Tim. 256 = p. 265 ed. Waszink (London & Leyden 1962). 3) Cf. J. H. Waszink, Die sogenannte Fünfteilung dey Träume bei Chalcidius

und ihre Quellen, Mnemosyne 1119 (1941), 65 f.; Blum, op. cit., 59. 4) Op. cit., 58.

402

Mras 1) has a different view: sonsniuns = 6veipoq (sic) ; spectaculum visum = revelatio = The ad-

monitio he does not even mention.

Switalski 2) has again another identification, but only for three

out of the five classes mentioned: yevelatio = somnium

s?ectaculum= The most recent effort is that made by Behr 3) : somnium =

visum - admonitio = 6veLpoq; sfiectaculu5Y1 = while Behr himself notices that revelatio bears no

relation to

If we put these four identifications 4) together, we see that only about one category there exists a communis opinio, namely somnium.

It can be put on a level with that is, if we forgive Mras his

apparent slip of the pen in identifying somnium with Eveipoq. It is clear that all four start from the presumption that the

classification made by Philo, the Hebraica philosoPhia, is connected

with the one by Artemidorus/Macrobius. That this presumption is

false I have shown above. All these so very divergent views are

only the logical results of it. Only Switalski has noted that in

Philo no trace of Calcidius' classification can be found, but Blum 5) thinks this of no major importance, because in his opinion all these

classifications ultimately go back to Posidonius. And after all, we

have only part of Philo's treatise on dreams.

Before going further into the classification of Calcidius some

remarks should be made on what Calcidius understands by the

Platonicum dogma in this connection. For that purpose we must

go back a few pages in his commentary. In ch. 253-255 he has

expounded his view more clearly. In brief it comes to this.

Plato 6) studied the problem of dreams and finally came to the

1) K. Mras, Macrobius' Kommentar zu Ciceros Somnium, Sitz. ber. Preuss. Akad. Berlin, Ph. Hist. K1. (1933), 237.

2) B. W. Switalski, Des Chalcidius Kommentay zu Platos Timaeus (Münster 1902), 45.

3) Op. cit. 177, note 11 f. 4) The interpretation of Waszink in Mnemosyne 111 9 (1941), 65 ff. will be

reviewed later on. 5) Op. cit., 59. 6) This and what follows of course is the interpretation of Calcidius.

403

conclusion that there are more than one source from which dreams

may originate 1). In the first place there are the dreams that are the results of

physical or psychical causes: gigni quibpe vel ex reliquiis cogitationum et accidentis alicuius rei stimulo, ut in nono Politiae libro docet 2).

Secondly there are the dreams that result from divine providence, or (and this is a third class) they result from the love that the

heavenly powers have for human beings 3) : at veyo somnioyum, quae divina pyovidentia vel 4) caelestium potestatum amore iuxta homines

oboyiuntur, causam rationemque exposuit in eo libro qui Philosophus inscyibitur 5). The examples that Calcidius gives for this expla- nation are from the Cyito (44 a I o-b 2) and the Phaedo (60 e).

A fourth class comes about when Plato reflects on the fact that

even in a waking state the human being is not without the bene-

volent help of God, and by this support is moderated in his ac-

tions 6) : denique etiam vigilanti non deerat proPitia divinitas, quae eiusdem modeyaretuy actus, ut Plato demonstyat in Euthydemo 7). As

an example he mentions the daemonion of Socrates. If we take a closer look at these definitions, we see that they

correspond respectively to somnium, visum, admonitio, and specta- culum in ch. 256 8).

But we expect to find a fifth class, too. It must be a fairly general one, if we consider Calcidius' definition of revelatio to which it

1) Comm. in Tim. 253. 2) Cf. Plato, Rep. 571 b 4-572 b 1, of which lines Calcidius l.c. gives a

translation. 3) Comm. in Tim. 254. 4) Vel = aut is not uncommon in post-classical Latin; cf. e.g. Tac. Hist.

1 21 oblivione apud posteros vel gloria distingui ; Tac. Ann. 14, 35. That vel should be taken here in this sense will appear in the course of my argument (cf. below, note 8).

5) By Philosophus he means Epinomis, viz. 984 e 5 ff. 6) Comm. in Tim. 255. 7) Here Calcidius errs; it is not in the Euthydemus but in the Theages 128

d 2-7. 8) This correspondence is immediately clear for somnium and spectaculum.

That the Hebraic visum and admonitio find their corresponding parts in the Platonic doctrine appears from cap. 254. Here we find divina providentia set against caelestium potestatum amore, while under the definition of visum we find the divina virtus as the cause of such dreams, and under admonitio the angelicae bonitatis consilia as the cause.

404

should correspond. I think it can be found in the words 1) : Nec veyo dubitaye fas est intellegibilem deum pro bonitate natuyae suae

rebus omnibus consulentem o_pem generi hominum, quod nulla esset sibi cum corpore conciliatio, divinayum Potestatum interpositione ferye voluisse; quayum quidem beneficia satis claya sunt ex pyodigits et

divinatione vel noctuyna somnioyum vel dittynafama byaescia yumoyes ventilante, medelis quoque adveysum ntoybos intimatis et

pyophetayum insplyatione veyidica.

On the details of the system as a whole the following obser-

vations may be made.

The definition of somnium given by Calcidius refers to a rather

general category of mostly non-predictive dreams. One can find

corresponding remarks in a wide range of authors. It may be iden-

tified with the ÈVÚ7tVWV of Artemidorus, but, as will be shown, it

is not the result of the same system. Next Calcidius gives a definition of visum : quod ex divina viytute

legatur. Again he traces the cause of the dream, why the soul sees

this particular dream. There seems, however, to be a difficulty in

the translation and exact meaning of the words. Waszink 2) trans-

lates : "das aus der gottlichen Kraft (nl. der Seele...) gewonnen wird". Accordingly, he supplies -animae and takes legatur as sub-

junctive from legere. He states: "Dafur spricht auch die Gegen-

überstellung von sornnium und visum durch quidem...... veyo". It is quite true that quidem ..... veyo can give a closer connection

to the words to which these particles are attached. But there is

no reason at all here to supply animae. There would be, if Calcidius had written: ... quod ex reliquiis commotionum animae q u i d em ....

followed by .... quod ex divina viytute verso ... etc.

To my mind another interpretation is more plausible here.

It may not be untrue that MMS can be translated on occasion

by 'Kraft', but it seems that here the meaning 'might' or 'om-

nipotence' would fit better 3). The word is often used for the divine

omnipotence 4). Of course it need not have the usual Christian

1) Comm. in Tim. 255. 2) Mnemos. 1119 (1941), 68, note 2. 3) Cf. Blaise & Chirat, s.v. nr. 5. 4) Cf. e.g. Tert. Prax. 26 virtus altissimi.

405

flavour with Calcidius, but it is remarkable that Calcidius speaks of a divin.a virtus. And let us remember that Calcidius probably was a Christian himself. The mental suppletion of animae, I think, is rather far-fetched. And furthermore it is improbable that legatur should be taken as a subjunctive. All around there are only in-

dicatives, also in the same construction. Therefore legatur is the

indicative from legare. And finally, with an eye on cap. 254 where

the words divina providentia clearly have a connection with Cal-

cidius' definition of visum, the passage should be translated as

'what is sent by (virtue of) the divine omnipotence'. So the purely

godsent dream i) is meant, the 'o'-veLpo4 of Philo.

The definition of admonitio does not in any respect resemble one

of the definitions of Macrobius. As to the contents, as they appear in the examples Calcidius gives, it might be compared to Macrobius'

oyaculum.

Spectaculum is the fourth class with Calcidius 2). It appears from

his definition that by it he meant the daemonion of Socrates. This

daemonion restrained, as we know, Socrates from doing things that were not good. So Calcidius has prohibens. Although he cites

as an example Theages 128 d 2-7, in which the task of the daemo-

nion is clearly defined, Calcidius does not excusively mean the

daemonion by his definition; he also uses iubens aliquid. Anyhow he wanted to make clear the auditive character of this class.

Revelatio forms the last class. As already noticed this is a rather

1) Plato did in fact know of the belief that dreams might be godsent; cf. Apol. 33 C

2) Spectaculum in this sense is of course highly unusual. Perhaps this is a translation of some Greek word. I know of only one passage where it is used in a somewhat similar sense, viz. Augustine in Psalm. 96, 1 magna spectacula Deus praebet cordi christiano. I am indebted to Mr. P. G. van Wees for this reference. In the Corpus Glossariorum Latinorum, rec. G. Goetz, it is said to be the equivalent of (II 505, 42 and III 338, 40), (III 302, 23 and 522, 13) or (II 532, 20). For used in connection with dreams cf. Arist. Div. Somn. 463 b 17

In 2 Mac. 15, II-16 we find the word used to describe a vision in a dream; cf. further Jos. Vita 208 Ant. II 10 and II 75; A cta Joannis 48

406

general class, only saying that the secrets of the future will be

opened in these dreams.

One would do better to give up attempts to connect the systems of Calcidius and Macrobius. This Waszink 1) already noticed

referring to Calcidius' words consentit huic Platonico dogmati Hebraica philosoPhia. But he did not ask himself why Macrobius'

system did not show any relation to that of Calcidius. We have

seen that this is not surprising, for Macrobius' system fundament-

ally differed from the Hebraicaphilosophia ( = Philo) by starting from different points of view.

The way in which the system reached Calcidius has already been explained by Waszink 2) : from Calcidius it goes back to

Porphyry, who got it from Philo via Numenius.

But if Calcidius' version of the Platonic system is in accordance

with the Hebraica philosoPhia, it must be possible to determine

the various identical classes.

Calcidius makes mention of five classes. Therefore it is certain

that the Hebraica Philosophia also knew five classes. In the part of

Philo's treatise that has come down to us only three classes have

been defined. The other two possibly had been treated in the part lost after the second book.

We further see that only the somnium in Calcidius seems to have

had no connections with the divine, but the other four classes

clearly do have a divine component. This shows that Calcidius drew a sharp distinction between

predictive (i.e. divine) and non-predictive (i.e. not divine, be-

longing to the human soul) dreams, apparently following the

Hebraica philosoPhia in this matter. This implies that we have to

search also for a class of non-predictive dreams in Philo.

However, all three classes mentioned by Philo have connections

with the divine, they are 0el<ey<«oi ovwpot. Therefore I think

that Waszink 3) is quite right when he remarks: "Was weiter die

erste Traumart, das ex reliquiis commotionum animae entstehende

1) Mnemos. III 9 (1941), 73. 2) Ibid., 81-82, and more recently in Porphyre, Entretiens de la Fon-

dation Hardt XII, 35 ff., espec. 53-54. 3) Mnemos. III 9 (1941), 79.

407

somnium betrifft, so kommt es mir sehr wahrscheinlich vor, dass

Philo sie in dem jetzt verlorenen Teil seines Werkes erortert hat". He then gives some good parallels for his statement that Philo did

not regard all dreams as godsent 1). But did Calcidius' somnium 2) concern only the non-predictive

dreams, as Waszink seems to imply when he draws a parallel with

Aristotle's De lnsomniis 46o b 28 ff. ? Plato 3) did in fact consider

some dreams to be products of so-called 'Tagesreste', to which he

also reckoned the effects of food and drinking. He therefore re-

commended as a means to exclude such effects a sober meal before

one goes to sleep (thus calming down the and urged the

reader to behave so that the will not effect

the dreams. Then the Xoyt.aTt.xo? can have its own way, and then

that part of the soul is able to foresee the future. So dreams may or

may not be prophetic, in Plato's interpretation. I wonder whether Calcidius did not identify this sort of dreams

with the ones which Philo described in his third class. This would

mean that Calcidius' somnium includes both Philo's third class and the non-predictive dreams which Philo occasionally mentions.

It may be useful here to stress the fact that Calcidius was in a

very difficult position when he tried to reconcile such divergent

systems as those of Plato and Philo, the more so as Plato strictly spoken did not have any regular system at all. Moreover, as Waszink

thinks, Calcidius may have had only a second-hand knowledge of

the classification-system in the Hebyaica philosoPhia. We further must bear in mind that Calcidius had to adapt the

1) Ibid., 81-82. According to LSJ, s.v. Artemidorus and Philo made a distinction between 'a mere dream' and 'a signifi- cant, prophetic dream'. But in Artemidorus this distinction is not always kept in mind (cf. above, p. 392, note i), and neither is it in Philo; cf. e.g. De Cherub. 128

De Somniis II 78 (by which Joseph is meant). But generally speaking Philo did make such a distinction, as appears most clearly from De Somniis II 138

etc. 2) Mnemos. 9 (1941), 83, he states that somnia are "solche, die aus den

von aussen her empfangenen Eindrücken hervorgehen", which is somewhat different from Calcidius' definition.

3) Rep. 571 b ff. Calcidius gives the whole passage in translation.

408

Posidonian system, in Philo's version, in accordance with the

passages in Plato about dream-like occurrences. He, as it were,

interpolated an existing dream-classification into Plato's theories.

Most important is the fact that Posidonius and Philo attached a

deorum adpulsus to (predictive) dreams, while Plato, of course, did no such thing.

So the definition of somnium which Calcidius gives in fact includes

both predictive and non-predictive dreams, but the cause of it

lies within the soul itself. Now Philo's 0el<ey<«oi 6veipoi are identical with visum in Cal-

cidius. Here I disagree with Waszink, because of his interpretation of ex divina virtute legatur, as I have explained above 1).

Philo's second class corresponds to Calcidius' admonitio, as

Waszink agrees 2). This class originates from cooperation with

the souls in the air, as Cicero puts it. In a more christianized

version these souls may become angels or divine spirits 3). As I said, Waszink identifies the third class of Philo with visum

in Calcidius. I may add to my criticism of this interpretation that

Philo did not credit the soul with a divina virtus or something like it,

only with a prognostic power, that is, in connection with dreams. The consequence of Waszink's interpretation is that he has to

rank both spectaculum and revelatio under Philo's first class, the

dreams sent by the divine. But it must be considered unlikely that

the only difference between the two is "dass die revelatio der

klarere Traum ist" 4). Philo does not mention any difference in

clearness for his dreams of the first class. He only draws attention

to the mutual difference in clearness of his various classes 5). The fundamental difference between yevelatio and spectaculum

is, however, that spectaculum clearly is a vision received in a

waking state 6), and revelatio any revelation of the future, especially of course in a dream.

By sbectaculum Calcidius understands the daemonion of Socrates,

1) Cf. above, pp. 404-405. 2) Op. cit., 79. 3) Philo De Somn. I 141. 4) Op. cit., 83, note 1. 5) De Somn. II 3-4. 6) Cf. below, p. 409, note 5 and p. 411, note 1.

409

which comes about by means of a heavenly power. So it bears some

resemblance to Philo's first class, but only a little. The fact remains

that Philo, in the part of his treatise on dreams which has come

down to us, did not speak about visions that arise when one is awake.

In my opinion, spectaculum must be put on a level with in the

Septuagint. Philo may have mentioned this class in the part of

his work that has been lost. In fact, the is a quite common

phenomenon in the Septuagint. As to the revelatio I may repeat what Waszink said : "dass die

nur den sofort verstandlichen Traum bedeuten kann, der Urheber nicht genannt wird" 1). It is remarkable, however, that Calcidius

did not give any example of it from Plato's works. We may suppose that if there were any, Calcidius would have mentioned them.

Furthermore this has been said to be the only time in non-

Christian literature that revelatio is used 2). However, Calcidius

seems to have been a Christian himself 3). The word is a translation, made perhaps by Tertullian 4) himself, of the Greek Now ('x7roxo'cXu?L? means both 'vision' and 'revelation'. So it is not

exclusively used for a certain class of dreams, but in any case it

can be used for dreams with a revelational character 5).

1) Op. cit., 83. 2) Cf. C. Mohrmann, Die altchristliche Sondersprache in den Sermones des

hl. Augustin I (repr. Amsterdam 1965), 144. 3) Cf. Plato Latinus IV, Timaeus a Calcidio translatus, ed. Waszink

(London & Leyden 1962), XI. 4) Cf. S. Teeuwen, Sprachlicher Bedeutungswandel bei Tertullian (Pader-

born 1926), 18; C. Mohrmann, op. cit., 80-81. 5) Cf. Andr. Caes. Apoc. 1

... , etc. Of course one would have preferred an older testimony than this one from the sixth century. Perhaps a passage in Clement of Alexandria will do, although he makes no explicit mention of dreams, viz. Strom. I 29 (p. III, I St.) :

Furthermore cf. 2 Cor. 12, 1 where the Vulgate translates visiones et revelationes. More pertinent is Hermas Vis. III 1, 2 (Joly) ....

etc.; cf. III 10, 7; Lxx Da. 2, 19 (Theod.) cf. Da. 2, 28,

where in both versions is used. One cannot help feeling that there was no clear distinction drawn be-

410

Therefore we must suppose that Calcidius, not finding anything in Plato resembling the Old-Testamentary decided to

adopt the explanation already provided for it, without bothering much about Plato. Moreover, his words imminentis exitus may point to the usual, i.e. principally eschatological, character of

in the Old Testament 1). As Calcidius actually was a Christian, this

easily explains why he shows some knowledge about the Old Testament. Another possibility is that Philo treated the

in the lost part of his work on dreams.

The result of these explanations must be the rejection of the

structure of the five-fold system, and its relations with Philo, as

Waszink sees it. In my own interpretation of the correspondence of the various

classes I have laid more stress on the fact that Calcidius was

starting from Plato's ideas, and therefore ran into difficulties when

applying the Hebraic system to the remarks on dreams that he

found in Plato.

Concluding this section I shall resume what I think are the

connections between Calcidius' and Philo's definitions in their

respective classifications.

Calcidius' somnium includes Philo's third class in addition to his

general remarks about non-predictive dreams (which probably form a separate class) as originated by the human soul.

Calcidius' visum coincides with Philo's 0el<ey<«oi that is

his first class.

Calcidius' adnzonitio is to be identified with Philo's second class.

Calcidius' spectaculum has very little connection with Philo's

first class (only because of caelestis Potestas). We have seen that

tween a vision of a person in a waking state and a vision in a dream. The fact that, according to Calcidius, the Hebraica philosophia ranked them both under a general classification of dreams is, I think, sufficient proof for it.

1) Cf. Religion in Geschichte und Gegenwart, I, col. 465, s.v. Apokalyptik : "Die Offenbarungen werden gern in der Form von Träumen und Visionen gegeben.....". For exitus in this sense cf. Tert. Resurr. 25 (incidentally, a chapter on the Apocalypse of S. John) in exitu saeculi ; Tert. Apol. 18, 2; Tert. De Monog. 16 illo ultimo exitu saeculi; Commod. Apol. 136 nullo datur scire, donec fiat exitus aevi.

Imminens corroborates, I think, my suggestion to take exitus in this sense; cf. e.g. Tac. Hist. 1 86, 3 omen imminentium cladium.

411

Socrates' daemonion is meant by it. It may be a translation of the

Greek (or oF«w5) which occurs in the Septuagint in connec-

tion with dreams. It is therefore the class in which a personal con-

tact between the dreamer and his god is established 1). Calcidius' revelatio, as we have seen, corresponds to the apo-

calyptic revelations in the Old Testament, which often have an

eschatological character, both in dreams and otherwise.

We know that what is left of Philo's work on dreams in only a

part of the whole work. One theory is that it originally consisted of

five parts, with only two of them remaining. However this may be, Philo may have reviewed in the part that has been lost other

dream-stories from the Old Testament, among them possibly the

and I must admit that this is rather 'spe- culative, but to me it seems the only possible way to connect

Calcidius' views of the Platonicum dogma with the Hebraica philo-

sophia.

VI

Something remains to be said about the way in which the five-

fold classification reached Macrobius. In short, who was his

source for Comm. I 3 ? The various views have been collected by W. H. Stahl 2).

'

Mras assumed Porphyry's Quaestiones Homeyicae as a source, while Courcelle had a preference for Porphyry's Commentary on

Plato's Republic. We may add that Schedler 3) connects the classifi-

cation with a commentary on the Timaeus, also by Porphyry, but

1) In Greece such a personal contact between god and man might be achieved by visiting an incubation-sanctuary. This explains the close re- semblance of Calcidius' definition of spectaculum and Macrobius' remarks on oraculum. Calcidius, however, applies the spectaculum to people who are awake (vigilantibus), while Macrobius explicitly adds in somniis. But in incubation-sanctuaries such contacts between god and man were often achieved in the twilight zone between sleeping and waking. Nevertheless spectaculum is included by Calcidius in a dream-classification (multiformis ergo est ratio somniorum, cap. 206), although it happens to people who are awake. See further above, p. 409, note 5.

2) Macrobius' Commentary on the Dream of Scipio, transl. by W. H. Stahl (New York 1952), 34.

3) P. M. Schedler, Die Philosophie des Macrobius und ihr Einfluss auf die Wissenschaft des Mittelalters (Münster 1916),85, note 6.

412

Courcelle thinks this improbable because of the terminology used by Calcidius.

Now there is a point in Courcelle's view that a commentary on the Republic may have been the source, for Macrobius cites the

dream of Agamemnon from Iliad II, a passage which is also treated

by Plato in his Republic. Courcelle further is quite sure that Ar- temidorus was not Macrobius' source 1). However, this does not

explain the almost literal resemblance between Artemidorus' and

Macrobius' remarks on the classification and explanation of

dreams 2). On the other hand, there is truth in Courcelle's remark 3) that

Macrobius drew a parallel between Plato's Republic and the dream

of Scipio in Cicero's De Re Publica. It is also true that for this pur-

pose he must have consulted the existing commentaries, especially the one by Porphyry. This shows the probability of Macrobius

having drawn from this source when he treats of the dream of

Agamemnon (Coww. I 3, 15 ff.). Blum 4) held that Porphyry had

been Macrobius' source by referring to auctoye PorPhyrio in Comm. I 3, 17. But this remark clearly refers only to the imitation by Virgil of the motif of the Gates of Dreams which we find in Homer.

But, although Porphyry had quite a big influence upon Macrobius, I fail to see why he must have been the source for Comm. I 3,

1-13, too. In order to find the possible source we first shall have to review

Comm. I 3 in connection with the surrounding chapters. As I see

it, this chapter is clearly an insertion, only serving as a learned

excursion. This view harmonizes quite well with Courcelle's words 5), "Macrobe, s'il n'est pas un compilateur, n'est pas un plagiaire: il

ne transcrit pas une source unique, mais son ouvrage revele 1'6tendue

de ses lectures".

The passage has a rather abrupt start, it serves as an introduction

to the phenomenon of the dream, and just before the end of the

1) P. Courcelle, Les lettres grecques en occident (Paris 1948), 24, note 2. 2) Cf. Blum, Studies, 53-56; Stahl, op. cit., 87-88, note 1. 3) Op. cit., 23. 4) Op. cit., 56-57. 5) Op. cit., 21, agreeing with Mras and Henry.

413

chapter 1) the whole classification-system is applied to the dream

of Scipio. Then Macrobius draws on another source (see above) and in addition gives some remarks on the dream of Agamemnon. I

think, therefore, that Macrobius has found the whole system ready- made in some more or less obscure work, probably on the inter-

pretation of dreams, and then put it into his own work without

bothering much about the very loose connection it has with the

surrounding chapters. This view, I feel, is strengthened by the fact that nowhere else

in his work do we ever come across either visio, or oraculum, or

visum, or even insomnium. If we leave the whole chapter out, noth-

ing much has been lost of the argument of the Commentary. All

this would imply that Macrobius did not use Porphyry as a source

for Co?ram. 1 3, except for the last part (1 3, 15 ff.). But there is another way to arrive at the same conclusion.

Porphyry was very much occupied with philosophico-religious

questions. In this sphere also falls the question of dreams and

their origin. So we may assume that this subject had his interest 2), but as I see it, mainly from the philosophical point of view.

As I have already mentioned, Waszink 3) has shown that Cal-

cidius made use of the classification of Philo, which ultimately came from Posidonius, and that it reached Calcidius by way of

Numenius and Porphyry. So Porphyry made use of the philosoph- ical classification, which bears no resemblance to the practical classification used by Artemidorus and Macrobius.

But why did Macrobius not use Porphyry's system? I think there is a very practical reason. Macrobius wanted to write a commentary on Cicero's Dyeam of Scipio, that is, he wanted to interpret the

dream. And besides he wanted to give an air of 'Wissenschaftlich-

keit' to it. So he disregarded Porphyry for the moment, and found

an impressive dream-classification in some dream-book, the same

that has been Artemidorus' source. This hypothesis gains in plausi-

bility when we bear in mind that Macrobius mentioned not only the

1) Comm. 13, 12-13. 2) It can also be inferred from his Qu. ad Iliadem, 217, 9 (ed. Schräder),

where the dream-book of Artemon is mentioned. 3) Cf. above, p. 406, note 2.

414

five classes of dreams, but also the division into five (= genera in Macrobius). Now where else could he have found it except in a dream-book ? We may take it as certain that he did not use Ar-

temidorus' work directly, for that would leave unexplained the

question of the source of his definitions of visum, oyaculum, and

visio, and why he did not mention the distinction between theore-

matical and allegorical dreams.

VII

In the above I have not taken into account the medical theories

concerning the dream, for instance the classification of Herophilus. Diels, Dox. Gr. 416, gives the following text : 'Hp6<pt.Xo<; ov?ipwv

0eo<ly<«ouq XKT' yiv?66av, TovS 8? cpua?xouS

6u E ov To Tou 8E

fauyxpmym<ixolqf 3) <05 4) X<XT' si8wawv 7tpócr7t'!'ú)mv a (?ou?o?,?8a wç iocS ópwv'!'CùV 5) Èv

\)7tvcp This text is to be found in Plutarch's Placita V I, 2. Galen in his Historia Philoso?hiae 106 (= Diels, Dox. Gr. 640)

has drawn on the same source as Plutarch : oveiPwv

1) Diels conjectured from Galen's version. The manuscripts, however, read which Bernardakis wants to keep in his text. This is, however, immaterial to my present argument.

2) too, has been conjectured by Diels from Galen. Bernardakis sticks to the reading of the manuscripts, but I fail to see what this would mean.

3) is the reading of the manuscripts. Galen has

4) 'Ex has been bracketed by Diels, kept however by Bernardakis, who is right in doing so, if one considers that Galen's version has

5) is read by both Diels and Bernardakis, probably with an eye on Galen's The or of the manuscripts does not seem to make much sense.

6) Cf. the critical apparatus in Diel's Doxographi, 640, showing the confusion of the text.

415

As is usually assumed, this classification has nothing to do with

the two preceding ones 1). But quite recently C. A. Behr 2) has

expressed a very different opinion. So a few words have to be said

here about his theories.

In Behr's theory of a gradual development of the classifications

into two-, three-, four-, and five-fold systems, Herophilus' classifi-

cation forms the three-fold system. Posidonius then would have

added a fourth class and then "finally by the first century A.D.

under the influence of medical theories on disruptions caused by the digestive processes, the non-predictive dreams were also di-

vided into two classes". So we have five classes. But if this were true, whose medical theories had such an in-

fluence, and why did not Herophilus himself come to this con-

clusion ? And if medical theories really gave rise to a splitting up of

the non-predictive dreams into two separate classes, we at least

expect to find some traces of a medical origin in these two resulting classes. The disruptions caused by the digestive processes, however, were known long before the first century A.D. We find some

reference to this question in Plato's Republic C 3) , but it might easily be older. Certain prescriptions of the Pythagorean school 4) seem

to point in this direction. Their views on dreams Behr does not even

mention, although they had, as it were, a sort of ritual 5) before

going to sleep, a ritual which very appropriately has been brought into relation with their beliefs on dreams 6).

Besides, a systematic interpretation of dreams was known long before Artemidorus. Panyasis, for instance, who lived in the 4th

century B.C., has been credited with the division into 1).

1) Cf. Waszink in his commentary on Tertullian's De Anima (Amsterdam 1947), 502.

2) Aelius Aristides and the Sacred Tales (Amsterdam 1968), chapt. 8, espec. 174, note 11 ; cf. above p. 392 and notes 2 and 4.

3) 571c. 4) Cf. Waszink on Tertullian's De Anima 48, 3. 5) Iambl. V.P. 65 and 114. 6) Cf. P. Boyancé, Le culte des Muses chez les philosophes grecs (Paris

1936), 110. 7) Artem. II, 4-6 Pack. Did Hippocrates show any knowledge of the

division into in 87 Cf. the and in Artemidorus p. 7 Pack.

It is not really important to our argument, whether Panyasis in fact

416

Returning to Herophilus' classification I must admit that it seems quite obscure what exactly he means. But I do not think Behr's solution acceptable. He identifies the so-called cpuacxoi

ovsc?oc with "the predictive dreams of the soul" i). It seems to me

that, if Herophilus had meant such dreams, he would have called them rather §uxixoi than puaixoi. For cpumxóç this sense is attested nowhere else. Moreover the usual sense 'arising from natural (or physical) causes' fits in here quite well 2). So here we have dreams in which the dreamer, and therefore also his physician, might get some knowledge about his physical condition. The words

Èaó!LEVOV certainly seem to be in favour of this explanation. The same conception is to be found in Hippocrates' work on

dreams : 8È: TO xoci EYp7jYopeou(7<x

4). So the soul is credited with quite an independent

activity during sleep. It even can foresee certain physical abnor-

wrote about this whole system or only about public dreams, as del Corno, of. cit., 122 assumes. The important point is that the system was in existence in the fourth century B. C. If we are right in assuming a certain connection between this system and n. IV 87, this would involve no problems of dating, for it is not improbable that the Hippocratic work has been written about 370 B.C.

1) L. Edelstein, Ancient Medicine (Baltimore 1967), 242, goes even further: "Herophilus declares dreams to be merely psychological phenomena; it is not the bodily changes but only the psychic changes that are manifested in dreams; these are the natural dreams which have to be separated from the divine". However, he cites only half of Herophilus' theory on the Placita, and puts, I think, a modern theory into Herophilus' mouth.

2) So already C. Fredrich, Hippokratische Untersuchungen (Berlin 1899), 215: "im Zustande des Körpers .... beruhende Träume".

3) In Galen's version we read accepted it seems by E. R. Dodds, The Greeks and the Irrational (Berkeley 1951), 124, note 28. I must admit that

does not make much sense. Wyttenbach (IV 436) already wrote : "malim ". But why not The sense of

would be: 'what is going to happen in consequence of those dreams'. For this sense of cf. LSJ., s.v. C III 2, and Kühner-Gerth, Satzlehre, I, 520. Palaeographically the difference between and is very little. If we then suppose that had been abbreviated to, for in- stance, the following could have been misread as if was attracted by ou and are easily confused in minuscule hands.

4) II. 86; cf. Galen VI 834 ed. Kühn.

417

malities: 6xo'aoc «05 1). In the passage from the Placita Herophilus makes a clear dis-

tinction between godsent and non-godsent dreams:

puaixoi, dreams arising from a divine cause and dreams arising from natural causes, the latter being associated with the constitu-

tion of the dreamer, as we might expect from the physician He-

rophilus was. I think, therefore, that for instance Bfchsenschftz

was completely wrong in maintaining: "Die Beziehung auf k6r-

perliche Zustande, deren Berucksichtigung man von einem Artzte

erwarten sollte, fehlt" 2). We should remember that the physical state of the body can be foreseen by the soul, as Hippocrates

clearly stated for the first time. But we cannot call the class of the

cpu6cxov 6veipoi merely 'predictive dreams of the soul', for this is

only the interpretation of but they are dreams caused by natural causes, as contrasted with divine

causes telling the dreamer r6 auyp£pov xml r6

7tpOç Tourouq 3) So Herophilus' first class consists of dreams that are due to a

divine origin, whether we read or OeonvEUC7,rOU4. They

happen x«T' &v&yx'Y)v, which probably means that the human being has no part in it as to their origin; in other words, they are sent

from outside, from the gods. This class coincides with the OeZcx

E-vu7rvLoc of Hippocrates 4), who reserves the interpretation of this

category for the dream-interpreters. So we may take it as certain

that Herophilus followed Hippocrates in his description of these

two classes.

But what about the third class attributed to Herophilus ? A glance at Diels' Doxogyaphi will tell us that there is not even a

communis obinio on the exact wording of the passage. C. Fredrich

adopted a conjecture by Diels and read ?v?u?.«TCxouS 5). This

1) II. 87. This same opinion is held by Galen VI 833 Kühn, although he puts it in a negative form:

2) B. Büchsenschütz, Traum und Traumdeutung im Alterthume (1868, repr. Wiesbaden 1967), 34. He was followed by M. Gelzer, op. cit., 42.

3) Cf. above, p. 416, note 3. 4) II. 87. 5) Op. cit., 215.

418

conjecture at first seems attractive, if one looks at the passages which Diels and Fredrich adduce to support their view. However, this conjecture has been refuted in part by M. Gelzer, whose further

conclusions I shall not follow 1). In addition to this refutation it

may be remarked that the fundamental difference between dreams

originated by the soul and those originated by the human body, which Diels and Fredrich seem to presuppose, did not exist at all.

Bodily conditions might be seen by the soul in dreams, as Hippo- crates tells us 2).

The reading 'mixed together' gives rise to con-

siderable difficulties. Firstly, what has been mixed together ?

Secondly, how are we to explain ex To6 which seems to

mean almost the same as X (X'!" And thirdly, what about

X CXT I think that Dodds was on the right road, when he suggested

that the words oTav a pouxlye0m Tow

£v form a fourth class 3). Dodds further thinks that these "'mixed' dreams are dreams of monsters

which on Democritus' theory 4) arise from a fortuitous conjunction of He should have referred to the same chapter in the

Placita, where it is said: rouq lveipovq Y,LvzaoocL XOC'7o'c '!'(;)v d8NÀú)V napaa?aa?v. This rather disproves his theory, for

in Democritus' theory all dreams come about by way of a con-

1) Op. cit., 41 f. Gelzer sees the relationship between Herophilus and Artemidorus as follows: = =

and = The following objections can be raised: a) It is not clear why Artemidorus should have followed the medical tradition in his classification of dreams; b) Artemidorus has five classes, whereas Herophilus has only three; c) Artemidorus does not start from an aetiological reasoning, as Herophilus seems to do especially in his first and second classes; d) The parallels between Synesius and Artemidorus which Gelzer adduces to prove his theory, do not prove that their classifi- cation-systems were the same, but only that there are certain connections between Artemidorus and Synesius in their views on the aetiology of the dream.

2) II. 87; cf. also Herophilus' second class. 3) Op. cit., 124, note 28. 4) As Blum, op. cit., 69 already observed, might

be Democritean in origin; cf. Plut. Plac. 899 E and Plut. Conv. 735 B (sc.

419

junction of and certainly not all dreams were dreams of

monsters. 'Fortuitous' is Dodds' translation of casu in Lucr. 5, 741, but there is not a single corresponding word in the Greek text of the

Placita.

Futhermore, against Dodds' interpretation of the third class it

might be asked whether it needs have a Democritean origin, and

why all of a sudden this Democritean theory emerges in the work

of this medical writer. We further should bear in mind that e18mxov

does not have a Democritean flavour every time that the word

occurs, and that, notwithstanding the parallels which Blum 1) adduces, 7tpócr7t'!'ú)mç is not exclusively a Democritean word, and

neither is ?cPoa?ri?Twv 2). Moreover 7po'a7rrwgL4 is not the same as

7t(XprXcr'!'(Xmç. Therefore Dodds' "(fortuitous) conjunction of

may be a translation of but certainly not of 1tpócr7t'!'ú)mç. It has been suggested to me that this mixture to which auy-

refers, is a mixture of divine and human causes. Then

x«T' r?Poanic?aw would be the divine component and

pouxlye0m the human one. But as Dodds already noticed 3), "a dream of one's beloved is not a 'mixed' dream ....

in any sense". One also expects that, if auyxpmym<ixo4q were the

correct reading, the mixture would consist of 0el<ey<<oq and

cpvwxoS, but a pouxlye0m can hardly be called

The words ex Tou x«T' d8NÀú)V npo?nTC?aw have nothing to do with the divine, but rather suggest a mechanical automatism.

On the other hand, Artemidorus also mentions that are

common to both body and soul 4), and it is in just this connection that he mentions the examples of a lover being together with his

beloved boy in his dreams, and a sick man dreaming of being healed. If this remark has anything to do with Herophilus' classification,

1) Cf. the preceding note. 2) The verb for instance occurs twelve times in Hippocrates'

cf. R. Joly, Hippocrate, Du Régime (Paris 1967), index, s.v. 3) Op. cit., 124, note 28. This holds good as far as the categories 'divine'

and 'physical' are concerned. In Artem. 4, 2 we read about dreams that are a mixture of somatic and psychic elements, it seems.

Artem. 4, 4 Pack The text is somewhat confused; cf. below, p. 421. Reiske's conjecture is now confirmed by the recently discovered Arab translation of the first three books of Artemidorus; cf. Gnomon 37 (1965), 671. The passage seems to have a very medical colour.

420

it would give a clue as to the sense of then the

origins of such dreams would have been mixed together, that is to

say, these dreams would be products of both body and soul. How-

ever, the supposed mixture would consist of cpumxóç and

Body and soul as origins of such dreams then would not be in

accordance with these terms. So Artemidorus is better left out, at least for the moment.

But what if we read auyxp?KTLxouf; ? This in fact has already been done by M. Wellmann 1), and to my mind he has come up with some good reasons. He referred to Soranus 2) for evidence that

Herophilus used the word a4yxpiym especially to describe a part of

the body. To this may be added the use of the word by Galen and

Soranus in the sense of 'anatomical structure', to show that it was

well known in medical circles. Hence Wellmann concluded: "Die

6veipoi auyxpL?t(xTLxot sind also Traum, welche durch ein Organ des

Korpers hervorgerufen werden". The fact that the reading auyxpL?taTLxo'uq comes from Galen is

in its favour. Galen was the man who must have had a better know-

ledge of medical terms than the writer of the Placita. Besides, it is

more probable that someone changed auyxpiym<ixo4q, which he

did not understand, into a word the sense of which would be

immediately clear. It is also quite possible that the scribe had in

mind a three-fold classification of dreams, for instance the one by Posidonius and Cicero, and then incorrectly drew some parallels between the two systems 3) . And so he credited Herophilus as well

with a three-fold system, the third class of which being a sort of

combination of the other two. All this seems to me sufficient reason for keeping Galen's reading

cyuyxpL?taTLxouq. But then what of Tou As I have

already noticed, Èx rou seems to express almost the same

1) Über Träume, Arch. f. Gesch. d. Mediz. (1924), 72; cf. A. Palm, Studien zur Hippokratischen Schrift (Tiibingen 1933), 72, who adopts his conclusions.

2) Gynaec. 1131, 85 (p. 372, 13 ed. Rose). 3) Such parallels were in fact drawn by Wellmann, op. cit., 72. This of

course is wrong, for Herophilus did not follow Posidonius' classification, but adopted the views of Hippocrates on these matters, certainly so for his first and second classes.

421

thought as x«T' &vrXyx'Y)v. Some scholars therefore have treated it as a

gloss and excluded it from the text 1). But it is not quite the same

as x(x'!" ocvocyxyl having of itself already connections with

the divine, i.c. ex Tou certainly does not have

this colour. Here, therefore, it indicates a corporal automatism

showing itself to the dreamer's mind x«T' d8c0Àú)v 7rp6a7rrwatv. The

"3wXa occasioned by physical stimuli automatically reach the

mind of the dreamer. I think that by these observations we can meet

Dodds' criticism of Wellmann's reading 2). Now we are left with a pouxlye0m etc., for this

kind of dreams cannot be explained as corporal automatisms only. As I observed above, Dodds' suggestion that this might form a

fourth class, to me at least, seems attractive. So something must

have dropped out after 7tpócr7t'!'Cùmv 3). It may not then be pure coincidence that Galen's text is very confused precisely in these

last lines. I suggest that after 7tpócr7t'!'ú)mv we should read ro'uq 8E §uxixo4q 4).

For this seemingly wild conjecture a parallel is to be found in

Artemidorus: 8E Tou',7cov rcov & 8c' £v8eimv

K 8e , ?rr?pcasoi? , ?« 'RCOV 8' K plpov oc. 8'e ,

at' 5). It is rather remarkable that only a few lines before, Artemidorus speaks about lovers dreaming of their beloved boys s),

although the text seems to be somewhat confused. Furthermore, this chapter in Artemidorus has an undeniably medical colour 7).

1) So Diels, Dox. Gr., 416, followed by Blum, op. cit., 69. 2) Op. cit., 124, note 28. 3) Cf. the critical apparatus in Diels' Dox. Gr., 640. 4) It is not the word that is of most importance, but rather the

idea of a separate class of dreams, for which we can find parallels in other medical writers.

5) 4, 7-9 Pack. 6) 4, 2-3 Pack; cf. 3, 17-18 Pack; Macr. Comm. 1 3, 4 si amator deliciis

suis aut fruentem se videat aut carentem ; Schol. on Ar. Nub. 16. Perhaps this kind of dreams was a common topic in the medical schools. Such dreams were already mentioned by Aristotle Insomn. 460 b 5-8, who of course treated them quite differently.

7) Cf. Artem. 239, 2-4 cf. Artem. 4, 5-7 ;

Hipp. 93, 2-3. Galen attaches importance only to the somatic cause, leaving the psychological one out; cf. VI 834 Kiihn

422

Speculative though my conjecture may seem, if we accept the

insertion of ro'uq 8s we have a classification of dreams which shows a good deal of relationship to the theories put forward first by Hippocrates, and later on by Galen. And this certainly is

something we might expect in Herophilus 1). A short recapitulation shows the following results. First He-

rophilus draws a distinction between and cpuavxoi 6veipoi, that is, as to the origin of the dream. His theory is that they come

about either because of some divine influence or by certain natural

causes, that is, when the soul is active in the body and forms mental

images. So this first part is an answer to the vexed question of how

do dreams arise.

Then Herophilus elaborates the second possibility, viz. the kind

of dreams that were important to him as a physician. He then asks

himself: what do these dreams result from or by what are they caused, in short, why do we have such dreams ? He finds two

causes, namely a physiological one and a psychological one; such

dreams are the result of either psychological or somatic troubles 2). Until now I have only hinted at Galen's views on the dream. I am

fairly sure that his ideas must have some relationship to the theory ascribed to Herophilus. Therefore I quote the most important

passage on dreams in his works in full 3) : 8E EV TOGS oux

To6 rdvzw cpaviaaia xmplq To6 ml>mX«a0«1 7rCkVU X(x6(X7rSp x«1 To6 £a01cw iTCX?CF'7(Oq Toiq 7tZLV&m, TOU 8È 7oZq O'7tÉp[1.iXTOe; r?a?peaw. It may be noticed that in each of the passages adduced there is talk of shortness and abundance of food and drink on the one hand, and of fear and hope (or desire) on the other hand.

1) Cf. Dignot. ex Insomn. (VI 832-835 Kühn). A large part of De Dignatione has been repeated in Galen's commentary on Hippocrates' De Humoribus II 2 (Kühn XVI 219-226). The more theoretical part, however, has been left out.

2) In this way Hippocrates distinguishes between dreams caused by wishes of the soul 93, 3), and e.g. those caused by want of food 93, 2).

It is rather surprising, that Dodds, op. cit., 124, note 28, after his sugges- tion of a fourth class arising from did not think of the pos- sibility that the third class might arise from bodily disorders.

3) Kiihn VI 833.

423

The first thing to notice is that Galen does not consider all

dreams to be of value for the diagnosis of his patients. He clearly allows room for mantic dreams. So we again have a division into

medical and non-medical dreams, which Herophilus also made, with an important modification, however. For Galen does not

credit the gods with the sending of prophetic dreams, but considers

this, too, to be a faculty of the human soul 1). The words cw 7tECPPOV'!'[X(X!LEV bear resemblance to the a

pouxlye0m of Herophilus: both clauses express kinds of

wish-fulfilment-dreams. So Galen also gives an answer here to the

question zvky dreams occur. Another possibility, then, is that dreams are the result of our

daily occupations, thoughts of which occupy us even in our sleep 2). But that is not all. Galen clearly states that because of these

various sorts of dreams it becomes difficult to give a diagnosis of

the body by virtue of the drams arising from that very body. Now at first sight dreams that arise from the body seem to be

the only sort there is. However, a few lines further on Galen im-

plicitly seems to make a sharp distinction between the purely mantic dreams and those arising from the body 3). It may be con-

cluded that he did not hold mantic dreams to come directly from

the body. This, then, brings his theory closer to the classification

of Herophilus: Galen has mantic dreams, without explaining how

exactly they arise, only stating that it is the soul that displays a

mantic activity, whereas Herophilus has Eveipoi. But neither the kind of dreams that results from our daily occu-

pations, nor the one that results from what we continually think

about, has any significance for diagnosing the body. Galen says:

1) So in his opinion the only way in which dreams can arise, is through the working of the soul, either or in the sphere of the bodily con- stitution. As he adds, it is often very difficult to choose between the two.

2) Cf. Aristotle's theory in his work on dreams, Div. Somn. 463 a 25; and Hippocrates 88, 1-2.

3) (VI 833 Kühn).

424

easy for Galen, if he only had to make a distinction between

'Tagesreste' -dreams and medical ones, to consider those dreams the

result of a bodily constitution that showed no connections with our

daily waking-life. His point is that the mantic dreams make such a

distinction extremely difficult. However, it seems to be clear that

Galen wishes to distinguish also between purely somatic (that is,

being of value for the diagnosis) dreams, and dreams that result

from so-called 'Tagesreste'. In fact this is the same distinction that

Herophilus made, namely between cruyxpc?,«T?xo? and what I

exempli gratia called

By way of summary it may be said that there is a great deal of

resemblance between Herophilus' and Galen's conceptions of the

dream. Both distinguished between godsent, c.q. mantic, dreams

and non-mantic dreams. The last class fell within the competence of the physicians. This sort of dreams arises from natural causes, which can be divided into psychological and somatic causes 2).

1) VI 833 (Kühn). 2) I thank Dr. K. J. McKay from Melbourne, who during his recent visit

to this country found some spare time to turn my English into a more in- telligible form. Prof. Verdenius and Prof. Waszink have read this paper in typescript and made some valuable suggestions, for which I am grateful.

UTRECHT, Vleutenseweg 163