

Carawan


————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 2011 Edwin Carawan 
 
 
Paragraphê and the Merits 

Edwin Carawan 

On Archinus’ suggestion you enacted a law that if anyone 
should bring a suit contrary to the oaths, the defendant could 
make a special plea; the magistrates would introduce this issue 
first and the man who brought the special plea would speak first. 
Whoever lost would pay a one-sixth penalty, so that those who 
dared to recall past wrongs (mnêsikakein) would not only be con-
victed as perjurers, but would also be penalized immediately, 
without awaiting punishment from the gods.1 

OT LONG after democracy was restored, Archinus in-
troduced a special remedy against lawsuits that violated 
the oaths and covenants of the Amnesty. This pro-

cedure for paragraphê was to be one of the defining reforms of 
the new regime; inspired by the reconciliation agreement and 
broadly adapted in later law, no other procedure is better 
represented in the extant speeches. The distinctive features are 
described here in the prologue to Isocrates’ speech Against 
Callimachus: the archon will introduce this issue first and the 
defendant will speak first, challenging the suit against him; the 
loser will owe a penalty of one-sixth the amount at issue. So it is 
that Isocrates’ client speaks first, as he is the defendant: he 
 

1 Isoc. 18.2–3: εἰπόντος Ἀρϱχίνου νόµον ἔθεσθε, ἄν τις δικϰάζηται παρϱὰ 
τοὺς ὅρϱκϰους, ἐξεῖναι τῷ φεύγοντι παρϱαγρϱάψασθαι, τοὺς δ’ ἄρϱχοντας περϱὶ 
τούτου πρϱῶτον εἰσάγειν, λέγειν δὲ πρϱότερϱον τὸν παρϱαγρϱαψάµενον, ὁπό-
τερϱος δ’ ἂν ἡττηθῇ, τὴν ἐπωβελίαν ὀφείλειν, ἵν’ οἱ τολµῶντες µνησικϰακϰεῖν 
µὴ µόνον ἐπιορϱκϰοῦντες ἐξελέγχοιντο µηδὲ τὴν παρϱὰ τῶν θεῶν τιµωρϱίαν 
ὑποµένοιεν ἀλλὰ κϰαὶ παρϱαχρϱῆµα ζηµιοῖντο. Transl. Mirhady, in David C. 
Mirhady and Yun Lee Too, Isocrates I (Austin 2000) 98. Elsewhere trans-
lations are my own except where noted. In this study “the Amnesty” refers 
to the diallagai of 403. 

N 


 EDWIN CARAWAN 255 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
challenges the plaintiff’s claim (10,000 drachmas) for money 
that was confiscated under the oligarchy, on the grounds that it 
violates the Amnesty. If the paragraphê goes against him, Cal-
limachus will owe one-sixth of that sum. In that respect, the 
roles are reversed: the defendant becomes plaintiff.2 

This testimony is open to interpretation and the other evi-
dence is often equivocal, but it was concluded long ago that the 
new procedure involves a two-stage trial.3 That the archons 
shall “introduce this issue first” seems to imply that there is first 
a hearing to determine whether the suit is admissible, and then, 
if the challenge is rejected, the court takes up the original com-
plaint.4 Whoever loses on the procedural issue would owe the 
epôbelia, whatever the outcome on the main suit; for, sup-
posedly, the paragraphê requires a separate decision.  

The later cases refer to statutes barring lawsuits for pro-
cedural errors, as in Roman exceptiones,5 and it has been 
supposed that paragraphê was available in all such cases as, for 
instance, when a suit was brought in the wrong court (exceptio 
fori). Such is the plea opposed by Lysias 23 Against Pancleon (the 

 
2 Thus R. Dareste, Les Plaidoyers civils de Démosthène (Paris 1875) xx: “Le 

défendeur qui opposait la paragraphè devenait demandeur non pas seule-
ment aux fins de son exception, mais pour tout le letige. Il parlait le premier 
sur la fin de non-recevoir d’abord, et ensuite sur le fond, car la question du 
fond n’était pas réservée, et il fallait toujours plaider à toutes fins. Les rôles 
des parties se trouvaient ainsi complétement renversés, à ce point que le 
reject de la paragraphè entrainait contre celui qui l’avait opposée condam-
nation à l’épobélie.” Cf. Dem. 34.4, κϰατηγορϱεῖν τοῦ διώκϰοντος. 

3 This was already assumed by M. H. E. Meier and G. F. Schömann, Der 
attische Process (Halle 1824) 645–647; followed by J. Lipsius, Das attische Recht 
und Rechtsverfahren III (Leipzig 1915) 846. By their view the epo ̂belia is assessed 
(at least in the later cases) if either party fails to win one-fifth of the votes.  

4 Largely from the late lexicographers, Lipsius, Recht III 857 n.39, con-
cluded that both paragraphê and diamartyria led to “delayed judgments,” 
ἀναβόλιµοι δίκϰαι. To the contrary, W. Hellebrand, “Παρϱαγρϱαφή,” RE 18 
(1949) 1169–1181, at 1176.  

5 Or praescriptiones; but see U. E. Paoli, Studi sul processo attico (Padua 1933) 
120, against the parallel. 


256 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
only instance roughly contemporary with Isoc. 18): there the 
defendant’s maneuver is called antigraphê and there is little 
similarity to the later pleas called paragraphai; yet scholars have 
often assumed that it is a variation on Archinus’ model. The 
procedure was evolving, but the basis for it was, as Wilamowitz 
described it, just such “form- oder competenzfrage.”6  

On this reckoning the procedure that Archinus introduced 
soon after 403/2 was an ad hoc solution to a singular problem,7 
and this technical solution opened the door to other procedural 
objections in the later paragraphai. Initially this recourse was 
based on the claim that the suit was in violation of a general 
amnesty (as it is usually interpreted), and thereafter it could be 
adapted to unrelated objections, precisely because it was 
originally and essentially a procedural remedy divorced from 
“the merits.”  

The idea that such questions should be decided separately 
from the main issue is a mainstain of legal thinking, of course, 
in both the Roman and common-law traditions. In Roman law 
procedural exceptions were regularly decided in the prelim-
inary hearing before the magistrate, in jure. In the American 
system, for instance, an appellate court may reconsider ques-
tions of law but refuse to reconsider the lower court’s verdict on 
factual issues; for the jury’s sovereignty as “trier of fact” is en-
shrined in the Bill of Rights.8 So it seemed reasonable to sup-
 

6 Aristoteles und Athen (Berlin 1893) II 368–373. Wilamowitz sees the time 
constraint as indicating a second stage of the trial: “offenbar hatte er wenig 
wasser, weil diese vorfrage erst von der eigentlichen abgetrennt worden 
war” (369). We return to this problematic case at 259–261 below. 

7 For the date of Archinus’ reform see D. Whitehead, “Athenian Laws 
and Lawsuits in the Late Fifth Century B.C.,” MusHelv 59 (2002) 71–96; 
followed (for the most part) by E. Carawan, “The Athenian Law of Agree-
ment,” GRBS 46 (2006) 339–374. 

8 The rule derives from the Seventh Amendment: “In suits at common 
law, where the value in controversy shall exceed twenty dollars, the right of 
trial by jury shall be preserved, and no fact tried by a jury shall be otherwise 
reexamined by any court in the United States than according to the rules of 
the common law.” 


 EDWIN CARAWAN 257 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
pose that a comparable division of issues emerged at Athens.  
But then Paoli presented an alternative construction:9 the 

real issue for the Athenian jury to decide, in civil suits where 
paragraphê applies, is whether to authorize the plaintiff to carry 
out his claim (demand payment or seize assets in lieu of pay-
ment), and the legalities cannot be severed from that reckoning. 
So there was no two-stage trial, with a separate hearing on the 
procedural question. Instead, the plaintiff’s claim and the de-
fendant’s challenge are introduced in the same hearing (the 
latter first); and, if the paragraphê is rejected, the same jury pro-
ceeds directly to a second ballot on the original complaint.10 
That unified procedure respects the principle of “inscindibi-
lità”: there is no divide between the substantive rights and the 
process to exercise them. After all, the extant paragraphai show 
that the plaintiffs argued their case on the merits, and even the 
defendant, though insisting that plaintiff’s claims be barred, de-
voted much of his argument to answering their claims.  

To illustrate Paoli’s model: in Isoc. 18 the issue is whether 
Callimachus will be authorized to demand payment or seize 
 

9 Paoli, Processo 77–116; followed by Hellebrand, RE 18 (1949) 1178–
1179, and A. Biscardi, “Giudizi paragrafici,” in A. Azura and E. Eula (eds.), 
Novissimo Digesto Italiano VII (Turin 1961) 879–880. Also skeptical of a 
second hearing: Dareste, Plaidoyers xx (quoted n.2 above); J. Miles, “Some 
Observations on Demosthenes’ Speech against Pantaenetus,” Hermathena 85 
(1955) 50–66. Cf. A. Steinwenter’s review of Paoli, ZRG 54 (1934) 382–387, 
suggesting that the protocol varied and evolved over time (esp. 385). 

10 In his conclusions Paoli outlines the sequence as follows (Processo 113–
114): “Quando il convenuto avesse prescelto questo mezzo de difesa [i.e., 
paragraphê], ne derivavano delle notevoli conseguenze procedurali. Il prin-
cipio che reus in excipiendo actor fit influiva anche sulla forma del giudizio in 
quanto:  1) il convenuto aveva per primo la parola, e lasciava perciò 
all’attore quel vantaggio, che tutte le legislazioni antiche e moderne 
attribuiscono al convenuto o all’accusato, di poter chiudere il dibattito;  2) 
se l’eccezione non riportava il quinto dei voti, il convenuto era colpito da 
un’ammenda, ἐπωβελία … Il magistrato che aveva la direzione del dibattito 
… doveva, terminata l’orazione del convenuto e dell’attore, sottoporre alla 
votazione del giudici anzitutto la questione paragrafica, quindi, a meno che 
l’eccezione non fosse stata accolta, l’ἔγκϰληµα dell’ attore.” 


258 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
assets, if Isocrates’ client will not settle; and our unnamed de-
fendant (NN) has a good deal to say against the merits of 
Callimachus’ claim, insisting that he is lying about what hap-
pened. He expects Callimachus to respond by arguing his case 
on the facts at issue in the hearing at hand. So, if the paragraphê 
is rejected, there is really nothing left to argue over; the jury 
will cast a second ballot, now on the original complaint. 

Wolff set out to disprove Paoli’s theory and largely suc-
ceeded.11 But Wolff himself was not so categorical as his 
followers have been. Some of his findings may tell conclusively 
against Paoli but not so unequivocally in favor of his own 
construction. On balance, it seems reasonable to reject Paoli’s 
double verdict, but the old two-stage trial remains precarious. 

Wolff’s argument against Paoli turns on two disparate find-
ings. (1) There is no evidence or clear indication of the second 
ballot; that much is indisputable. But (2) Wolff argues that the 
extant speeches are in fact well-focused on the procedural issue, 
the paragraphê itself; and that finding is not so conclusive. Most 
readers have been struck by how much the speeches labor the 
merits, but Wolff discounts that impression as “an optical il-
lusion.”12 In that regard, we should at least give the speeches 
another reading. And in his finding against the double verdict, 
Wolff himself scrupulously acknowledged, neither is there any 
clear reference to a second hearing, if the paragraphê is rejected. 

Following those implications, this essay offers a new model: 
(§1) In paragraphai there is no second stage to the trial—neither 

 
11 H. J. Wolff, Die attische Paragraphe: Ein Beitrag zum Problem der Auflockerung 

archaischer Prozeßformen (Weimar 1966). Cf. A. R. W. Harrison, The Law of 
Athens II (Oxford 1971) 108–119; D. M. MacDowell, The Law in Classical 
Athens (Ithaca 1978) 215, concluding “Paoli’s view “has been decisively 
refuted by Wolff’s detailed study”; S. C. Todd The Shape of Athenian Law 
(Oxford 1993) 138, “Paoli’s hypothesis … was demolished by Wolff.” Wolff 
himself showed admirable caution in his conclusions (84–85), even ac-
knowledging the value of “inscindibilità” in a diminished sense.  

12 E.g., Paragraphe 17 n.1: “dieser Eindruck … das Ergebnis einer op-
tischen Täuschung war…” 


 EDWIN CARAWAN 259 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
Paoli’s second ballot nor Wolff’s second hearing. Each invoked 
the silence of our sources against the rival theory, but the 
evidence weighs against them both. And (§2) the litigants 
themselves are not so keen on the divide between fact and for-
mality as later rhetoricians would be.13 From the first example 
to the last, the paragraphai represented in the speeches involve a 
peculiarly contractual principle. In the case against Callima-
chus, the question for the jurors to decide is whether earlier 
agreements (including the Amnesty of 403) have preempted the 
claim. Similar issues predominate in the later instances: after 
an arbitrated settlement or binding agreement, is the plaintiff’s 
claim foreclosed or already fulfilled in accordance with the 
covenants?14 This is not to deny that other, purely technical 
objections were also acknowledged, as paragraphai came to be 
more widely applied. But, in the cases documented by the 
speeches, the issues that emerge from the claim and the chal-
lenge lend themselves to a single decision. If the paragraphe is 
rejected, the plaintiff’s claim is affirmed and he will proceed to 
demand payment or seize assets in compensation. After all, 
Isocrates’ client, in explaining what appears to be an unfamiliar 
procedure, never says that the jury must reach a separate decision 
first on the paragraphê, only that this is introduced first and the de-
fendant will speak first. 

But let us first try to gain some perspective on Pancleon’s 
case (Lys. 23), as it is often treated as the missing link in the 
two-trial model. The speech tells us little or nothing about the 
original complaint but, aside from that silence, there is no sign 

 
13 The scholarly division of issues seems better suited to the later rhetorica 

than to the fourth-century realities; cf. E. Carawan, “What the Laws Have 
Prejudged: Παρϱαγρϱαφή and Early Issue-Theory,” in C. Wooten (ed.), The 
Orator in Action and Theory in Greece and Rome (Leiden 2001) 17–51. The entry 
in Pollux 8.57, in particular, relies on later rhetorical hypotheses (on fic-
titious issues). 

14 Paoli supposed, “very likely, in all cases where ‘the obligation was 
extinct’ it was possible to pose an exception by paragraphê” (Processo 95–96). 


260 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
of a separate hearing on the merits.15 To be sure, the text we 
have addresses only the procedural issue,16 answering Pan-
cleon’s claim, that the suit cannot be tried in the polemarch’s 
court because he has citizen rights; but that does not mean that 
this was all the plaintiff had to say in the hearing at hand. The 
tale of his damages, whatever wrong Pancleon has done him, 
may be so straightforward as to need no speechwriter’s art; the 
text begins with the sort of formula that might often suffice, 
when a speaker turns from his narrative to a complicated 
issue.17 So our plaintiff probably consulted the speechwriter 
only in regard to the more complicated question raised by 
Pancleon’s antigraphê.18 And on that issue he concludes his argu-
ment with a telling parallel (13–14): when Aristodicus brought 
suit against Pancleon, the defendant raised the same objection, 
that he could not be sued as an alien because he was (as he 
claimed) a Plataean. In that instance the objection was quashed 
by sworn testimony (diamartyria), without jury trial. In that 
parallel case Pancleon’s plea is simply called his antomosia (his 
formal response to the charge), and that may be all it is in this 

 
15 Cf. Edward Cohen’s reply to S. C. Todd, “Status and Contract in 

Fourth-Century Athens,” in G. Thür (ed.), Symposion 1993 (Cologne 1994) 
146–149. 

16 The formula µὴ εἰσαγώγιµον εἶναι does not always indicate a para-
graphê (for trial): see 274–276 below; recall that Callimachus’ claim had been 
quashed by a diamartyria, sworn testimony that the case was not admissible 
because of a prior settlement: οὐκϰ εἰσαγώγιµος ἦν ἡ δίκϰη διαίτης γεγενη-
µένης (11). This was before Archinus introduced the new procedure; so, if 
there were no witness willing to swear, the defendant would have had to 
argue that point at trial on the main claim (just as Antiph. 5 argues abuse of 
procedure). On the sense of eisagein/eisago ̂gimos, see at nn.60, 71 below. 

17 πολλὰ µὲν λέγειν, ὦ ἄνδρϱες δικϰασταί, περϱὶ τουτουὶ τοῦ πρϱάγµατος 
οὔτ’ ἂν δυναίµην οὔτε µοι δοκϰεῖ δεῖν. The deictic τουτουὶ suggests that he 
has at least summarized his grievances. For similar paraleipsis cf. Dem. 23.90 
(περϱὶ αὐτοῦ τούτου πολλὰ µὲν λέγειν οὐκϰ οἶµαι δεῖν), 40.38, 18.50; Isoc. 
20.153. 

18 K. J. Dover, Lysias and the Corpus Lysiacum (Berkeley 1968) 165, suggests 
that Lys. 23 is a client copy, showing the client’s own alterations. 


 EDWIN CARAWAN 261 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
instance—antigraphê usually means no more than the text of the 
counterclaim.19 The case at hand has come before a jury be-
cause (apparently) Lysias’ client has not found anyone willing 
to deny under oath that Pancleon is a citizen (to decide the case 
as Aristodicus did, by diamartyria). But there is nothing to sug-
gest that Pancleon’s plea is now any different. And, aside from 
the missing narrative, there is nothing to suggest that this case 
will be given a separate hearing on the merits, when Pancleon’s 
antigraphê is rejected. Indeed the speaker concludes with the 
usual injunction to the jury, as though the verdict in this hear-
ing will decide the case, not just the procedural issue (16): οἶδ’ 
ὅτι τά τε δίκϰαια κϰαὶ τἀληθῆ ψηφιεῖσθε, ἃ κϰαὶ ἐγὼ ὑµῶν δέοµαι.  

From these considerations I conclude that Pancleon’s case 
does not tell us much about early paragraphai. Whatever it may 
say about evolving procedure, it does not refer to a separate 
hearing on the merits. 

Whatever we make of the case against Pancleon, the other 
instances are very different disputes. In each example, from the 
case against Callimachus down to the latest of the mercantile 
suits, there is a tangle of questions that must be argued ab initio: 
Are the claims to contract or legal settlement fair or fraudu-
lent? And where do we draw the line between the rights 
recognized in the agreement and further claims that were not 
clearly addressed?  
1. A hearing on the merits 

Wolff’s scenario proceeds as follows: 
Of course, introducing a paragraphê delays the debate on the 
merits of the case and leads immediately to a preliminary that is 
independent, though the main issue remains pending. In pro-
cedural terms, this means that, if the paragraphê is rejected, there 
is no need to initiate the case anew, but, rather, the magistrate 
now brings that issue before the court. The question whether 

 
19 Cf. Dem. 45.45–46, reporting the antigraphai in the suit for false testi-

mony against Stephanus. Pancleon’s case would be the only instance where 
a paragraphê is called antigraphê. 


262 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
this sequel was before the same jury that decided the prelim-
inary or a newly allotted panel, would be of no consequence; it 
depends on whether there is enough time remaining, after the 
preliminary concludes, to bring the main issue to a conclusion 
before the same judges, on the same day.20 

This picture involves some awkward complications, both with 
regard to the practical arrangements and in the juristic 
rationale. Athenian justice seems to have followed the principle 
that trials begin and end on the same day—even in cases of life 
and death (Pl. Ap. 37C).21 But if the second hearing proceeds in 
the same court on the same day, there may be some confusion 
when the tables turn.22 These complications can be managed, 
of course, but it is surprising that there is no mention of them. 
Let us begin with what our sources do not say and then pro-
ceed to the recourse they recognize.  

As Wolff emphasized, “nowhere, not even in the speeches 

 
20 Paragraphe 84–85 (the emphasis is his): “Zwar verhinderte die Einlegung 

einer Paragraphe einstweilen die Einführung des Meritums und führte zu-
nächst zur Instruktion eines selbständigen Vorprozesses, doch hob sie die 
Anhängigkeit der Hauptsache nicht auf. Prozeßrechtlich ausgedrückt heißt 
das, daß es nach Zurückweisung der Paragraphe keiner erneuten Einbringung 
(λαγχάνειν) der δίκϰη bedurfte, sondern der Magistrat nun ihre Einführung 
beim Dikasterion bewirkte. Ob dies dann dasselbe Dikastenkollegium war, 
das über den Vorprozeß entschieden hatte, oder ein neu ausgelostes, wird 
ohne Belang gewesen sein und davon abgehangen haben, ob nach Beendi-
gung des Vorprozesses noch genügend Zeit übrig war, um die Hauptsache 
am selben Tage vor den gleichen Richtern zu Ende zu führen.”  

21 Ian Worthington has argued for the exception in cases such as the 
apophasis against Demosthenes in 323 (and he may be right): “The Length of 
an Athenian Public Trial: a Reply to Professor MacDowell,” Hermes 131 
(2003) 364–371. But such exceptions hardly disprove the rule: even for cap-
ital cases, the trial should ordinarily conclude in a day. In private suits we 
have no sign of an exception. 

22 From one vote to the next, the jurors will assign the opposite meaning 
to their ballots: in the paragraphê, the first urn or pierced ballot will be for the 
defendant, the second urn or solid ballot for the plaintiff; then at the main 
hearing, presumably, the plaintiff would speak first, so the pierced ballot is 
in his favor (on ballot arrangements see Todd, Shape 132–133). 


 EDWIN CARAWAN 263 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
opposing paragraphai, is there any mention at all of the need to 
renew the suit”; he takes that silence as a sign that, if the 
paragraphê were rejected, the court would proceed directly to a 
hearing on the merits.23 Presumably the procedural sequence 
was well known, and so, supposedly, a decision against the 
defendant on the paragraphê would lead directly to a trial on the 
original complaint, without any further formalities and without 
comment. Yet, if there is to be a separate hearing on the merits 
—and the jurors will not quite be done with the case when they 
reject the paragraphê—we might expect plaintiffs to say some-
thing to that effect, somewhere in the corpus of speeches on 
paragraphai. The speakers (on both sides) never say anything of 
the sort. Wolff admitted as much, as the same silence tells 
against Paoli’s model;24 but his followers have not been satisfied 
with that disclaimer or the uncertainty about when the main 
hearing would come. 

Thus MacDowell supposed that the hearing on the main 
issue would have to be decided “at a later date” and he found 
this indicated in the prologue to Dem. 36, For Phormio: τὴν µὲν 
οὖν παρϱαγρϱαφὴν ἐποιησάµεθα τῆς δίκϰης οὐχ ἵν᾽᾿ ἐκϰκϰρϱούοντες 
χρϱόνους ἐµποιῶµεν (2).25 Here, supposedly, “the speaker denies 
that the postponement of the trial … was his motive.” But that 
denial might simply mean that they did not introduce the 
paragraphê in order to derail the arbitration, where the suit be-
gan and might have reached a conclusion.26 After all, Phormio 

 
23 Paragraphe 85: “nirgends, auch nicht in den zur Abwehr von Paragra-

phai gehaltenen Reden, von der etwaigen Notwendigkeit einer neuerlichen 
Klagerhebung die Rede ist.” 

24 Paragraphe 83: “Spricht somit keine Quelle für grundsätzlich ungebro-
chene Kontinuität der Verhandlung zu Paragraphe und Meritum, so gilt 
allerdings das gleiche auch für das Gegenteil.”  

25 MacDowell, Law 215. Against “the usual view,” cf. Miles, Hermathena 
85 (1955) 64–65, finding it “absurd that the whole case should be re-tried 
after the parties had already argued fully on the merits and the facts … 
Surely the hearing of the plea would be regarded as a trial of the action.”  

26 On the effect of entering a paragraphê in arbitration see S. Isager and M. 
 

264 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
had reached a settlement with his litigious step-son on two 
previous occasions, and it was probably assumed that he would 
settle yet again. If the speaker had meant to disavow a delay 
within the proceedings at hand, he would probably have said 
so more plainly: “Our aim was not to postpone your verdict.”27 
Instead he speaks of prolonging the dispute more generally and 
with obvious irony; for endless litigation—on the part of Apol-
lodorus—is a major theme of the speech. 

If Wolff’s scenario holds true, we might especially expect the 
plaintiffs to say something about the second stage of the pro-
ceedings, where their claims would be given a proper hearing: 
“Put these technicalities aside and give us a fair trial—today!” 
The closest approximation is Dem. 35.43, where the plaintiff 
calls on the judges to “demand that [defendant Lacritus] show 
either that they did not receive the money, or that they repaid 
it, or that maritime contracts need not be valid, or that they 
should use the money in any other way than under the terms of 
the contract.” Now the defendant has already made the first 
speech, so it is fair to ask, when or how will he have the chance 
to respond? Wolff saw this injunction to the jury as a veiled de-
mand for them to reject the paragraphê and proceed to the main 
issue.28 But it seems out of character for this plaintiff, who is so 
emphatic about every detail of the proceedings, to speak so 
obliquely of his one chance of success. It seems more his style 
to demand that the jury voice their outrage directly (in thorubos), 

___ 
H. Hansen, Aspects of Athenian Society in the Fourth Century B.C. (Odense 1975) 
124–125. 

27 As Paoli remarked, Processo 107: “se avesse voluto accennare al 
differimento del giudizio principale avrebbe detto οὐχ ἵνα τὴν κϰρϱίσιν ἀνα-
βαλλώµεθα.” 

28 Paragraphe 78: “nichts anderes als das in rhetorische Form gekleidete 
Verlangen, dem Beklagten durch Zurückweisung seiner Paragraphe die 
direkte Verteidigung und damit das voraussichtlich hoffnungslose Suchen 
nach tragfähigen Gegenbeweisen aufzuzwingen.” Of course, the “veiled 
demand” may be simply a commonplace from contract cases where the 
plaintiff speaks first. 


 EDWIN CARAWAN 265 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
and let Lacritus try to answer; the aim is to catch him off 
guard.29 

Of course, in the corpus of eight speeches that seem fairly 
complete we have only two for the plaintiffs: Dem. 34 and 35.30 
But these seem largely representative and consistent with the 
defendants’ speeches in this respect: the litigants (on both sides) 
would try to exhaust the arguments they would use in any 
hearing on the merits. In fact we might have supposed that the 
plaintiffs’ speeches were actually written for hearings on the 
main issue, were it not for brief passages disposing of the plea 
to bar litigation. Thus in the speech Against Phormio (Dem. 34) 
the plaintiffs waste few words on the paragraphê (3–5) and then 
rely almost entirely on the testimony to what happened (only 
reverting to the paragraphê in conclusion, 43–45). The issue is 
whether the defendant (another Phormio) must pay off the pen-
alty under a contract he made to deliver a cargo to Athens; the 
defendant contends that the contract is no longer valid because 
the ship was lost at sea (and the contract recognizes that 
exigency); so he has invoked the law allowing paragraphê in cases 
 

29 The dismissive εὖ οἶδ’ ὅτι οὐδὲν ἂν τούτων οἷός τ’ εἴη οὗτος οὔτε 
διδάξαι οὔτε πεῖσαι (“I’m confident that he could neither explain or per-
suade (you) of anything”) suggests that AA expects to catch Lacritus with no 
answer (much as Plato’s Socrates catches Meletus, Ap. 34A–B). So this was 
probably a demand for spontaneous response. It is also possible that the 
litigants had a brief rebuttal after the first round of speeches (as the anon-
ymous referee suggested); that would match the sequence for ordinary dikai 
(Ath.Pol. 67.2). 

30 Neither does Hyperides(?) Against Demeas (P.Oxy. XXVII 2464) give any 
sign of a subsequent hearing on the merits, but it is too fragmentary to carry 
any weight. The only passage indicating a paragraphê is col. iii.11–24: AA 
protests the contradiction, if the law establishes guardians for orphans, to 
safeguard their interests, but the jury should decide that suit against the 
wrongdoing guardians be barred (µὴ εἰσαγώγιµον εἶναι ψηφίζ[ο]ισ̣θε). 
Here Wolff observes (Paragraphe 19), “Apparently all that the plaintiff aims to 
achieve is quashing the paragraphê; for him the main issue is a cura posterior.” 
But we hardly have enough of the connected argument to draw that conclu-
sion. In this essay I also avoid questions of authenticity, indicating by the 
attibution “Dem.” only that a speech belongs to the Demosthenic corpus.  


266 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
where there is no contractual obligation (symbolaion). In re-
sponse, the plaintiff introduces a barrage of testimony on every 
stage of the voyage (6–16), to show how Phormio has systemati-
cally cheated him of payment and security; and then a second 
speaker turns back to the sequence of litigation to discredit 
Phormio’s scenario and his chief witness, Lampis (who initially 
denied taking payment and then in arbitration admitted it). 
This line of argument takes up most of the rest of the speech, 
up to what appears to be a commonplace in such cases: If the 
case is rejected here, what recourse is there—“to what court 
shall we take our case, … if not to you (in the jurisdiction) 
where we made the contract?” (43). And, after one last attack 
on Lampis, the speaker concludes, “I have said all that I can. 
But I shall call another to speak in our behalf, if you require it.” 

Evidently the plaintiffs mean to say all that they have to say 
in the hearing at hand. If the jurors want to hear more, they 
will say so (in thorubos), and he will call the other speaker. If 
there were the chance of a second hearing, we would expect 
the litigant at least to allude to it: “If you have any reservations, 
reject the paragraphê and let us proceed to the main event; then 
we can answer any lingering doubts, and our opponent will 
have to answer our argument on the merits.” But he does not 
anticipate that option.  

In the parallel passage, Dem. 35.47–49, the plaintiff develops 
the commonplace in more detail: “Where must we turn for 
justice?” If not here, where maritime suits are decided, perhaps 
to the Eleven? But they deal with kakourgoi and capital crimes. 
Perhaps to the archon eponymos? But he deals with disputes over 
heiresses and orphans (etc.). Perhaps to the archon basileus? But 
he deals with asebeia (etc.). Perhaps to the polemarch or the 
generals? But these, again, have no jurisdiction over maritime 
contracts.  

The commonplace may have been inspired by the plea that 
the court has no authority.31 But it has developed into a broad 
 

31 Not precisely that the case is in the wrong jurisdiction but that there is 
no court competent to hear it—no magistrate can “bring in” the claim (or 
 

 EDWIN CARAWAN 267 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
response to paragraphai generally: If the remedy in this court is 
disallowed, where can we turn for justice? That development 
makes it all the more striking that, even after closing all the 
other options, the plaintiff never says, “Let us proceed to a 
proper debate on the merits.”  

All the paragraphê speakers—plaintiffs and defendants alike—
conclude in much the same way, as though the debate at hand is 
the only hearing on the merits.32 Of course the argument from 
silence cannot be conclusive, but it must weigh at least as 
heavily against Wolff’s model as against Paoli’s: if the absence 
of any mention tells against a second ballot, surely it tells 
against a second stage of the trial with another round of argu-
ments. And there are passages where the silence is especially 
persuasive, where the speaker turns to what recourse he has, if 
he loses the paragraphê hearing. 

The regular recourse is a suit against the witnesses.33 Such is 
the case of Apollodorus against Stephanus (Dem. 45). As we 
noted, Apollodorus had reopened some old claims against his 
guardian Phormio, going back to the disposition of his inheri-
tance, and this Phormio (not the plaintiff in Dem. 34) defeated 
the suit by insisting that the matter was closed by prior settle-
ment. Indeed, Demosthenes’ For Phormio presented a com-
pelling argument on those grounds. So Apollodorus recalls:34 
___ 
the claimant); cf. Wolff, Paragraphe 97–101. The law in Dem. 37.33 allows 
paragraphai against suits for which there are no magistrates (περϱὶ ὧν οὐκϰ 
εἰσὶν εἰσαγωγεῖς), but that objection was not the basis for the paragraphê (it 
was deleted from the affidavit); see n.45 below. 

32 E.g. Dem. 37.58–60, first focusing on the question of fact, to this effect: 
How could I have done you any wrong when I wasn’t even in town? Then 
defending the grounds for paragraphê: Even in homicide cases settlements are 
binding; the most dire prospect is that the jurors may do away with the 
ancient rule for final settlement. If there were a sequel, the defendant would 
anticipate it: Do not be misled to think that you can reject the paragraphê and 
do justice in the hearing that follows. 

33 Cf. Dem. 34.28, 31, and esp. 46–48, implying that a suit for false wit-
ness is his only recourse.  

34 Dem. 45.6: πρϱολαβὼν δέ µου ὥστε πρϱότερϱος λέγειν διὰ τὸ παρϱα-
 

268 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
As he spoke before me and took advantage, because it was a 
paragraphê hearing and was not going to trial on the main issue, he read 
these documents (releasing Phormio from further obligation) … 
and so affected the judges that they refused to hear any utter-
ance from us; so I was penalized with the epo ̂belia, denied a fair 
hearing and humiliated like no one else I know. 

The picture of Apollodorus baffled by thorubos may not be quite 
true to life, but the implication is plausible enough: he speaks as 
though a suit that is contested by paragraphê goes to trial on that 
basis and is not expected to proceed to a hearing on the main 
issue. Of course the articular infinitive, τὸ παρϱαγρϱαφὴν εἶναι 
κϰαὶ µὴ εὐθυδικϰίᾳ εἰσιέναι, may be a simplification, but the 
whole tenor of the passage reinforces the sense of it.35 For 
Apollodorus goes on to say (7) that he can only sympathize 
with the judges who ruled against him, because the lying wit-
nesses were so unscrupulous. If that jury had had the option of 
dismissing the paragraphê and giving him a hearing on the 
merits, surely Apollodorus would have alluded to that option, 
to make the most of the due process that was denied him. 

There is one other passage where defenders of the two-trial 
model might be tempted to find an allusion to some subsequent 
___ 
γρϱαφὴν εἶναι κϰαὶ µὴ εὐθυδικϰίᾳ εἰσιέναι, κϰαὶ ταῦτ’ ἀναγνοὺς κϰαὶ τἄλλ’ ὡς 
αὑτῷ συµφέρϱειν ἡγεῖτο ψευσάµενος, οὕτω διέθηκϰε τοὺς δικϰαστάς, ὥστε 
φωνὴν µηδ’ ἡντινοῦν ἐθέλειν ἀκϰούειν ἡµῶν· πρϱοσοφλὼν δὲ τὴν ἐπωβελίαν 
κϰαὶ οὐδὲ λόγου τυχεῖν ἀξιωθείς, ἀλλ’ ὑβρϱισθεὶς ὡς οὐκϰ οἶδ’ εἴ τις πώποτ’ 
ἄλλος ἀνθρϱώπων. Isoc. 18.39 suggests that the plaintiff could avoid the 
epôbelia by abandoning his case, but evidently Apollodorus at least tried to 
argue his claim, the jury voted, and he was saddled with the fine. 

35 Cf. Dem. 34.4: ἀλλ’ ὑπὲρϱ µὲν τῶν µὴ γενοµένων ὅλως συµβολαίων 
Ἀθήνησι µηδ’ εἰς τὸ Ἀθηναίων ἐµπόρϱιον παρϱαγρϱάφεσθαι δεδώκϰασιν, ἐὰν δέ 
τις γενέσθαι µὲν ὁµολογῇ, ἀµφισβητῇ δὲ ὡς πάντα πεποίηκϰεν τὰ συγκϰεί-
µενα, ἀπολογεῖσθαι κϰελεύουσιν εὐθυδικϰίαν εἰσιόντα, οὐ κϰατηγορϱεῖν τοῦ 
διώκϰοντος, “The laws have granted paragraphê in cases where there are no 
contractual obligations at Athens or for an Athenian market; if one admits 
that there is (a contract) but argues that he has done all that was agreed, (the 
laws) order him to proceed to euthydikia and make his defense there, not to 
accuse the plaintiff.” Again, the options, paragraphê and euthudikian eisienai, are 
opposite paths to the verdict, with no sign that one would lead to the other. 


 EDWIN CARAWAN 269 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
hearing: this is the remark by Isocrates’ client, at the end of a 
passage anticipating what Callimachus will say, “soon it will be 
possible to reply,” τάχ’ ἀντειπεῖν ἐξέσται.36 But scholars have 
usually treated this as a reference to further arguments in the 
same speech, and for good reason. This turn of the argument 
begins (35–36) with the protest that Callimachus will make 
(supposedly) at the injustice of being threatened with a fine 
under the democracy (the epôbelia) for money that was taken 
from him under the oligarchy. Against that complaint, our 
speaker says, “I think it will be easy to reply” (ῥᾴδιον ἀντει-
πεῖν). He then proceeds to offer various considerations against 
that complaint: all the democrats who came home from Pei-
raieus have suffered some loss, and yet none has resorted to 
litigation of this sort (38). Moreover, Callimachus has the op-
tion “even now, before making trial of your judgment, to drop 
the suit,” thus to be rid of the risk (39). So, if he speaks of the 
wrongs done to him under the oligarchy, the jury must demand 
that he show that the defendant—the man on trial—is the one 
who has taken the money (40). The men the jury must con-
demn are those who committed the wrongs, not the innocent 
(41). And so Isocrates concludes this section: πρϱὸς µὲν οὖν 
τούτους τοὺς λόγους κϰαὶ ταῦτ’ ἴσως ἀρϱκϰέσει κϰαὶ τάχ’ ἀντειπεῖν 
ἐξέσται. Mirhady translates: “In response to those arguments, 
then, this is perhaps sufficient, and it will be possible now to 
raise my own objections.” For this sentence introduces a 
further defense of the Amnesty (42–50), calling upon the jury to 
remember the conflict it delivered them from, when they cast 
their votes (45); after which Isocrates turns to discrediting the 
litigious tactics of Callimachus (51–57).  
2. Framing the issue 

Now let us weigh what the speeches do have to say about 
what is at issue in paragraphai, beginning with the first instance 
(a) and then proceeding to the later adaptations (b). Other 
 

36 Isoc. 18.41, as suggested by the anonymous referee. Cf. Mirhady’s 
translation, in Isocrates 105–106. 


270 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
readers were struck, as Paoli was, by how the speeches grapple 
with the merits of the case, the facts in dispute and the evidence 
on those claims. Wolff’s study was a tour de force in dispelling 
that impression. But let us reconsider the key arguments, each 
in its own context.  

(a) Thus against Callimachus (as we saw above), the defen-
dant calls for the jury to demand that the plaintiff address the 
matter at issue, to show that the defendant—the man on trial—
is the one has taken the money, “that I am the one who caused 
the damages he aims to recover.”37 As Paoli argued, that de-
mand strongly suggests that the facts at issue will be decided in 
the proceedings at hand. But, as Wolff insists, that point in the 
argument has to be read in its historical context. It comes in 
anticipation of claims that Callimachus will make linking NN 
to the oligarchs: “If he recalls what happened under the oli-
garchy, don’t let him make accusations against them, for 
crimes that no one will defend, but insist that he show that I 
have taken the money.” 

In Wolff’s view, this instruction to the jury is “nothing more 
than a rhetorical device, to remind the judges emphatically, 
that it is not the injustice at large in the era of the despotic 
regime but solely the role of the defendant invidually that is at 
issue” (83). He sees it as an appeal to a general amnesty and, in 
that light, should not be construed as an argument on the 
merits. After all (in his view), the other paragraphai are based on 
technical rules, separate from issues of substance. And the basis 
for paragraphê in this instance is the oath mê mnêsikakein that 
Isocrates invokes in the prologue: the new procedure is a way 
of enforcing a pledge against prosecuting the wrongs of civil 
conflict. The amnesty thus amounts to an ad hoc limitation, and 
challenging the lawsuit on this basis is typical of paragraphê, as a 
 

37 Isoc. 18.40: ἂν δ’ ἄρϱα µεµνῆται τῶν ἐπὶ τῆς ὀλιγαρϱχίας γεγενηµένων, 
ἀξιοῦτε αὐτὸν µὴ ’κϰείνων κϰατηγορϱεῖν, ὑπὲρϱ ὧν οὐδεὶς ἀπολογήσεται, ἀλλ’ 
ὡς  ἐγὼ  τὰ  χρϱήµατα  εἴληφα  διδάσκϰειν, περϱὶ οὗπερϱ ὑµᾶς δεῖ ψηφί-
ζεσθαι, µηδ’ ὡς αὐτός δεινὰ πέπονθεν ἀποφαίνειν, ἀλλ’ ὡς  ἐγὼ  πεποί-
ηκϰα  ἐξελέγχειν, παρϱ’ οὗπερϱ ἀξιοῖ τἀπολωλότα κϰοµίζεσθαι. 


 EDWIN CARAWAN 271 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
matter quite apart from the substantive claim.  
But mê mnêsikakein did not always mean “amnesty” in the 

usual sense.38 In the earliest documents the oath “not to recall 
wrong” seems to function as a closing upon the covenants of a 
treaty or private settlement. In casual usage, of course, it may 
suggest the moral burden of “forgiveness.” But where the 
pledge is invoked as a rule that can be enforced, it regularly conveys 
a bar against any further dispute on the matters resolved in the 
agreement.39 So, in the era after 404, it does not imply a bar 

 
38 Paoli, Processo 122–123: it is not until the second century B.C. that 

amnêstia gains currency as “an act of clemency by the sovereign power.” The 
Athenian Amnesty in particular, “differs juristically from the amnesty of our 
positive law: in form, because it is not an act of indulgence by sovereign 
power but a renunciation of the remedies allowed by law, by contractual 
agreement on the part of the members of the two hostile factions; in sub-
stance, because it also cancels private suits [whereas modern amnesty deals 
with criminal complaints]; in its processual function, because it does not take 
effect ipso iure but must be invoked by the accused or defendant as an ex-
ception, that the plaintiff’s claim is inadmissible; in its extent, as it does not 
extend ... to cases previously decided.” 

39 So argued E. Carawan, “The Athenian Amnesty and the Scrutiny of 
the Laws,” JHS 122 (2002) 1–23; on the contractual implications, GRBS 46 
(2006) 368–374; cf. R. Waterfield, Why Socrates Died (New York 2009) 132–
134, treating mê mnêsikakein as a rule of “no reprise.” For the conventional 
view (forgiveness by the victor): C. J. Joyce “The Athenian Amnesty and 
Scrutiny of 403,” CQ 58 (2008) 507–518; cf. my reply, CQ forthcoming. For 
the main comparanda see Astrid Dössel, Die Beilegung innerstaatlicher Konflikte 
in den griechischen Poleis (Frankfurt 2003). We may now add the well-preserved 
text of a settlement at Dikaia on the Thermaic gulf with Perdiccas III as 
witness and guarantor (364/3): E. Voutiras and K. Simanides, “Δικϰαιοπο-
λιτῶν Συναλλαγαί. Μία νέα επιγρϱαφή από τη Δίκϰαια,” in Ancient Macedonia 
VII (Thessaloniki 2007) 253–274. Here we have detailed arrangements for 
lawsuits, within a time-limit, concluding with an oath that includes the 
formulae (67–82) not to recall wrong “in word or deed” (κϰαὶ οὐ µνησι-
κϰακϰήσω οὐδενὶ οὔτε λόγωι οὔτε ἔρϱγωι), to honor the pledges they have ex-
changed (πιστώµατα = συνθῆκϰαι), and to abide by the city’s verdicts (ἔν τε 
ταῖς δίκϰαις αἷς ἐδίκϰασεν ἡ πόλις ἐµµενέω). For a deeper perspective on 
syngnômê and comparable terms, see now David Konstan, Before Forgiveness 
(Cambridge 2010). 


272 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
against all retributive actions absolutely. The settlement (dial-
lagai) included remedies for bloodshed and lost property. And it 
did not bar recriminations for past liabilities to the polis—those 
protections had to be enacted into law. A general immunity is 
not at all what Isocrates implies where he explains the rationale 
for his client’s paragraphê. 

The lawsuit by Callimachus (C) violates “the oaths and cov-
enants” in two ways: (1) C is prosecuting an accomplice, whose 
role amounted to (at most) “informing or denouncing,” where-
as the covenants expressly barred prosecution for such com-
plicity. Moreover (2) whatever claim C could make upon our 
defendant, he had settled in arbitration, and such decisions are 
rendered final by a rule that was evidently embraced in the 
covenants and promptly restated in statute. Of course, NN 
argues that C is lying about the facts, as well, and those matters 
of fact bear directly upon the legal issues: C probably argued 
that NN was the instigator, chiefly responsible for the confisca-
tion, and therefore was not protected by the covenant on in-
formants. And C would claim that NN is lying about the 
settlement: NN says that C would deny that there had been 
any settlement at all. However it was framed, that justification 
is disposed of in the first part of the argument (13–18) where 
NN argues that the claim is indeed subject to the settlement 
(however deficient his proof).40 

The first guarantee is presented as the threshold for the 
paragraphê: NN calls for a reading of the text (19) and then 
summarizes it (20), “the covenants expressly absolve (διαρϱρϱήδην 
ἀφιεισῶν) those who informed or denounced or did anything of 
this sort” (ἐνδείξαντας ἢ φήναντας ἢ τῶν ἄλλων τι τῶν τοιούτων 
πρϱάξαντας). This paraphrase suggests that the clause was 
broadly framed to encompass both those who initiated formal 
proceedings (endeixis and phasis) and those were merely accom-

 
40 The received text gives no indication of any testimony to confirm the 

fact of the settlement or the terms of it. Most editors assume a lacuna in §10, 
to allow for testimony by the arbitrator Nicomachus. 


 EDWIN CARAWAN 273 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
plices or accessories (as NN claims to be). Those who shared in 
the proceeds of confiscation could not be prosecuted for that 
complicity.  

How that rule applies is illustrated in §23: we are told that 
Thrasybulus and Anytus, though they know the men who listed 
their property for confiscation (apograpsantes) “nonetheless do 
not dare to bring suit against them or recall wrongs,” ὅµως οὐ 
τολµῶσιν αὐτοῖς δίκϰας λαγχάνειν, οὐδε µνησικϰακϰεῖν. By the 
usual interpretation, this refers to a general amnesty barring 
legal recourse for any wrongs committed in civil conflict. But it 
is clear that property rights remain valid and returnees are 
entitled to reclaim what is theirs. Citizens could take possession 
of their land and houses and recover at least some of their 
movable goods (whatever had not been sold). Thus Thra-
sybulus and Anytus recovered their real property and secured 
whatever assets they could find, at the expense of those who 
had taken possession. But even the champions of democracy 
had not prosecuted the apograpsantes, those who had listed prop-
erty for confiscation and received part of the proceeds.41 These 
accomplices are protected, as Isocrates’ client claims to be, by 
the covenant shielding “those who informed or denounced or 
did anything of this sort.”  

But the second count in the plea, that the matter is closed by 
an arbitrated settlement, is perhaps more vital to the argument 
because it serves to remind the jury (from common experience) 
of just what it means to settle past grievances by contractual 
agreement. This settlement was concluded as a diaita epi rhêtois. 
In this sort of settlement the so-called arbitrator witnesses the 
agreement and takes custody of the document; his duty is then 

 
41 As S. C. Todd recognized, Athenian Internal Politics 403–395 BC with 

Particular Reference to the Speeches of Lysias (diss. Cambridge 1985) 108. H. 
Kühn, “Die Amnestie von 403 v. Chr. im Reflex der 18. Isokrates-Rede,” 
WS N.F. 1 (1967) 31–73, at 64 n.47, accepted the rule regarding informants 
and denouncers but discounted it in this case. T. Loening, The Reconciliation 
Agreement of 403/2 B.C. in Athens (Stuttgart 1987) 56, discounted the rule; it 
was implicit in the amnesty. 


274 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
to decide any dispute over compliance, if one party contends 
that the other has violated the provisions for payment or other 
performance.42 This arrangement is a way of guaranteeing the 
finality of the settlement. It is that principle that is crucial to 
this case and to the reconciliation. In accepting that settlement, 
Callimachus has renounced his right to any further claim in 
that matter. To amplify that principle Isocrates extols the “cov-
enants” (synthêkai) and the contract for peace that was built 
upon them (§§24–34). For that principle of finality is the es-
sence of the pledge mê mnêsikakein that sealed the covenants of 
the agreement that we call the Amnesty.  

(b) The later speeches mention quite a number of grounds on 
which to bar litigation: these include a statutory time-limit (pro-
thesmia) and the rule against claims without jurisdiction.43 The 
speakers sometimes refer to these rules as though they were 
summarized in a general statute. Be that as it may, the grounds 
cited in later proceedings seem to lack any common denom-
inator other than procedural defect, and that feature has en-
couraged scholars to suppose that the paragraphê is divorced 
from the main dispute, to be decided quite separately.  

But it is doubtful whether all of those rules that declare suits 
inadmissible (µὴ εἰσαγώγιµον τὴν δίκϰην or δίκϰας µὴ εἶναι) 
might lead to a jury trial. While defendants in paragraphai often 
reinforce their arguments with limitations on time or jurisdic-
tion, these technicalities are never introduced as the principal 
grounds for the plea to bar the plaintiff’s claim. In the speeches, 
the defendants base their case on laws affirming specifically the 
right to invoke the paragraphê (not simply that the suit is inadmis-
sible); and those grounds for trial usually involve contractual 

 
42 We have one other instance in this era: Isoc. 17.19. Cf. A. Steinwenter, 

Die Streitbeendigung durch Urteil, Schiedsspruch und Vergleich nach griechischem Rechte 
(Munich 1925) 135–140.  

43 The various grounds to bar a lawsuit are reconstructed by Paoli, Pro-
cesso 85–95 (esp. omnibus measures indicated at Dem. 36.25 and 38.5); by 
Wolff, Paragraphe 90–105; Isager, Aspects 126–129.  


 EDWIN CARAWAN 275 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
agreement.44 In many instances, purely procedural barriers 
(such as time-limit or jurisdiction) might be addressed by the 
archon or the arbitrator and would not in themselves constitute 
an issue for the jury.45 Indeed, where those issues are brought 
up in court, they are often introduced with a disclaimer: the 
scrupulous defendant would not rely on such objections; the 
procedural violation is merely a mark against the plaintiff’s 
integrity. Thus Dem. 33.27 mentions the one-year limit for 
sureties, only to dismiss it, οὐκϰ ἰσχυρϱίζοµαι τῷ νόµῳ. The de-
fendants in Dem. 36.26 and 38.17–18 mention the five-year 
limit for suits against guardians only as a yardstick by which to 
measure the injustice, as (in each case) the suit comes twenty 
years after settlement with full release.46  

To judge from the cases represented in our speeches, the 
paragraphai that went to trial were based on the laws that bar 
litigation after “release and quittance” and those that deal with 
contracts for mining or maritime trade.47 To be clear: it may 
have been possible, in principle, to bring a paragraphê on the 

 
44 Cf. Carawan, GRBS 46 (2006) 351–358, with n.47 below. 
45 Thus in Lys. 17.5, we are told, the defendants got the suit against them 

barred as being in the wrong jurisdiction (διεγρϱάψαντο), evidently dismissed 
by the magistrates (without trial). In Dem. 37.33–34, where the defendant 
invokes the rule barring suits “for which there are no eisagogeis,” he explains 
that he originally included that in his plea, in preliminaries before the thes-
mothetai, but they erased it from the official formulation for trial. Similarly 
Steinwenter (ZRG 54 [1934] 382–387) emphasized the shifting scope of the 
archon’s discretion.  

46 Esp. 38.18, τοῦ νόµου λέγοντος ἄντικϰρϱυς, ἐὰν µὴ πέντ’ ἐτῶν δικϰά-
σωνται, µηκϰέτ’ εἶναι δίκϰην. οὐκϰοῦν ἐλάχοµεν, φαῖεν ἄν. κϰαὶ διελύσασθέ γε, 
ὥστ’ οὐκϰ εἰσὶν αὖθις ὑµῖν δίκϰαι (“But we did bring suit, they may say. Yes, 
but you settled that suit, so you have no recourse”). 

47 Aphesis and apallagê: Dem. 37.1 = 38.1. Maritime contracts: 32.1–2, 23–
24; 33.1–2, ἵνα µηδεὶς ἀδικϰῇ µηδένα τῶν ἐµπόρϱων εἰκϰῇ. τοῖς δὲ περϱὶ τῶν 
µὴ γενοµένων συµβολαίων εἰς κϰρϱίσιν κϰαθισταµένοις ἐπὶ τὴν παρϱαγρϱαφὴν 
κϰαταφεύγειν ἔδωκϰεν ὁ νόµος; 34.4, οἱ µὲν οὖν νόµοι … ὑπὲρϱ µὲν τῶν µὴ 
γενοµένων ὅλως συµβολαίων Ἀθήνησι µηδ’ εἰς τὸ Ἀθηναίων ἐµπόρϱιον πα-
ρϱαγρϱάφεσθαι δεδώκϰασιν. 


276 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
basis on any law that says simply, “such suits are not admis-
sible.” But among the speeches for trial there is no wide range 
of issues; the cases all revolve around the finality of binding de-
cisions. This principle encompasses the contractual settlement 
of private quarrels and the covenants of business contracts that 
foreclose any further dispute. It is beyond our scope to proceed 
through all the arguments in detail; here it should be sufficient 
to show how consistently the litigants embrace this principle in 
the other speeches, long after the plea Against Callimachus. 

Thus in the paragraphê against Apollodorus, For Phormio (as we 
saw above, 263–264), the challenge to the lawsuit is based upon 
earlier settlements that should have barred any further claim 
on those matters. In the speech Against Pantaenetus the defendant 
begins on the same note (Dem. 37.1), invoking the laws that 
grant paragraphê against cases where a plaintiff has given release 
and quittance and yet brings suit in the same matter. And the 
speech Against Nausimachus (Dem. 38) begins with nearly the 
same words: δεδωκϰότων … τῶν νόµων παρϱαγρϱάψασθαι περϱὶ ὧν 
ἄν τις ἀφεὶς κϰαὶ ἀπαλλάξας πάλιν δικϰάζηται. 

The contract cases (Dem. 32–35) develop a variation on that 
theme. The defendants invoke the law authorizing suits in the 
maritime court in matters “for which there are written 
contracts (syngraphai) and obligations (symbolaia)” based on that 
agreement; the law grants paragraphai against claims that are 
barred by provisions of the contract.48  

Wolff begins his analysis of the speeches with the case against 
Apaturius (Dem. 33), because “[i]t shows with particular clarity 
that the speaker is solely concerned with the question of 
admissibility”; but it is also “an instructive example of how the 

 
48 It was once supposed that this court heard cases based upon either writ-

ten contract or (other) legal obligation: thus L Gernet, Droit et société dans la 
Grèce ancienne (Paris 1955) 186–187; and that seems to be Wolff’s assumption, 
notably in regard to Dem. 32. Most commentators are now reasonably con-
vinced that this law for dikai emporikai effectively required a written contract, 
and the symbolaia at issue depend upon that agreement: e.g. Todd, in Sym-
posion 1993 136–137. 


 EDWIN CARAWAN 277 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
speaker, by convoluted tactics, sometimes slighting the juristic 
implications, was able to conceal his objective” (25–26). That 
is, the defendant is unconcerned with the main issue—he is not 
arguing against the claim as he would do in euthydikia—but in 
order to see this, we must extricate the real issue from the di-
versions. We shall follow Wolff's exegesis below, but let us first 
get a sense of what the defendant (NN) says the case is about. 

This dispute has a history going back two years. The de-
fendant tells it in detail, as though those events have some 
bearing on the decision at hand. In brief: a shipowner from 
Byzantium, Apaturius, relied upon his countryman Parmeno 
and our Athenian defendant to lend him a sum of money, 
putting up his ship as collateral; Apaturius then tried to ab-
scond with the ship, but Parmeno intervened, seized the slave 
crew. and stopped the ship. NN, in disgust, sold his share to a 
company of bankers. Then all parties reconciled: the original 
articles of agreement, the synthêkai, were destroyed; and the 
parties formalized their settlement with release and quittance 
(aphesis and apallagê)—“so that I would have no further business 
with [Apaturius] nor he with me” (12). He calls witnesses and 
emphasizes this conclusion: “Since then I have had no con-
tractual bond with him, whatever” (13, µετὰ ταῦτα τοίνυν ἐµοὶ 
µὲν οὔτε µεῖζον οὔτ᾽᾿ ἔλαττον πρϱὸς αὐτὸν συµβολαῖον γέγονεν).  

Of course that was not the end of it. Parmeno brought suit 
against Apaturius, for assault, because of the beating he got 
when the slaves were taken. Then Apaturius apparently evaded 
a challenge to decide the dispute by oath and instead brought a 
counter complaint against Parmeno (14). Thereupon they 
agreed to submit their dispute to arbitration and drew up cov-
enants governing the procedure (γρϱάψαντες συνθήκϰας).  

That document is the crux of the case at hand. As NN tells it, 
the covenants specified that the arbitrators were to deliver a 
verdict that would be binding upon the two parties, if the three 
arbitrators were unanimous or two concurred; NN was one of 
the three arbitrators. But Apaturius contends that one arbi-
trator, Aristocles was to decide the case; the other two were 
merely mediators (and NN was one of them). Now, in Apa-


278 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
turius’ version, NN was named as surety for Parmeno, and that 
is the basis for his claim against him.  

As NN tells it, the lead arbitrator was an impartial com-
patriot of the litigants, Phocritus; but when Apaturius realized 
that Phocritus was leaning against him, he insisted that the 
document of their agreement be entrusted to Aristocles—who 
was also his surety. He then claimed that Aristocles was sole ar-
bitrator, with full power to decide the case. But when Aristocles 
was called upon to prove this claim from the covenants, he said 
the document had been lost (18). After this impasse, the parties 
tried again to draft an agreement for arbitration, but each in-
sisted upon the arrangement most favorable to his cause. So, by 
NN's description, the effort to reach a settlement failed: the 
original covenants were now disputed and there was no docu-
ment by which to decide the dispute. Nonetheless, Aristocles 
asserted his authority and was on the point of pronouncing a 
verdict in favor of his man, Apaturius; but Parmeno confronted 
him and rejected a unilateral decision as a violation of their 
agreement. Soon thereafter, however, Parmeno left town to 
deal with a personal disaster (for his family had been devastated 
by an earthquake). Yet, though both of the other arbitrators 
disavowed the process, Aristocles proceeded to make an award 
of twenty minas to Apaturius. And, with Parmeno unavailable 
to contest it, there the matter remained for two years, until 
Apaturius decided to sue NN as surety for the delinquent Par-
meno. 

Now it may be helpful to summarize the sequence in even 
shorter compass. There was first a contract to pay off liens on 
the ship; this led to a dispute which was resolved in arbitration, 
with release and quittance. There were then cross claims for as-
sault, which led to a second agreement for arbitration; but that 
agreement was nullified when the document was lost and the 
parties could not agree on the arrangements. And yet NN is 
now sued as surety for a judgment that was rendered under 
that second, defective agreement. 

Without the document, the plaintiff relies on what witnesses 
will recall. And, as Wolff observed, the defendant apparently 


 EDWIN CARAWAN 279 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
has no witnesses present to confirm his side of the case, that he 
was not surety for Parmeno, that in fact Archippus had stepped 
into that role. Instead the “witnesses” that NN promised (22) 
turn out to be purely figurative. Thus he cites the time that 
passed as “witness”; for there is a statute of limitations barring 
any action against a surety more than a year after the judg-
ment. Of course, our defendant will not insist upon that limita-
tion, but it gives him an argument from probability: Apaturius 
was in Athens within that period and would not have missed 
the opportunity, if he had a legitimate claim. The only live tes-
timony in this regard is introduced to prove that Apaturius was 
indeed present in Athens the year before (26).  

In Wolff’s view, all the lengthy narrative and elaborate argu-
ment over events long past are merely a cloak for the frailty of 
our defendant’s case. The main claim is that NN is liable as 
surety, and Apaturius will have witnesses to that effect.49 By 
this reckoning, the proof that the first round of the quarrel was 
resolved with full release is irrelevant; and the ground for the 
paragraphê, the claim that NN cannot be sued in the maritime 
court because there is no contract, is a very precarious footing. 
Every turn of the argument reveals some new diversion con-
cealing the lack of evidence: Why did NN not call Archippus 
himself to testify that the latter was named surety?50 As for the 
missing document, Wolff suggests that “many jurors may have 
suspected that the text was stolen ... at Parmeno's urging!” In 
§29, NN seems to recognize the distinction between procedural 
and factual issues only to confuse them: if he had been named 
surety he would never have denied it, “for the arbitrator’s ver-
dict was not according to contract, so I would not be liable as 

 
49 Cf. Isager, Aspects 152, against Wolff’s theory of the symbolaion in this 

case. 
50 Wolff raises the point, Paragraphe 32: “Why does he not produce the 

supposed surety Archippus?”; if he were deceased, defendant would surely 
have invoked that excuse. But then, in an added n.25a, Wolff concedes that 
Archippus would probably have fled rather than face litigation.  


280 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
surety.” Thus, in Wolff’s view, he practically concedes that his 
argument on the facts is beside the point.  

Of course we should not make too much of the litigants’ nar-
rative.51 But we should also be wary of our own assumptions 
where they run counter to the text. If we assume that the 
paragraphê hearing is all about the procedural barrier, then, of 
course, all of the argument on the main issue is simply a 
strategy of evasion.52 To be sure, NN has an awkward gap in 
the evidence: the contract is missing and he has no witness to 
prove that he was not named surety. But I doubt that failing to 
prove the negative would be decisive. After all, is it really sur-
prising that Archippus is not anxious to step into the de-
fendant’s role and make himself liable? In this instance, just as 
in the other speeches, the defendant paragrapsamenos seems com-
pelled to argue on the merits as well as the procedural issues—
to prove both that the plaintiff’s suit is contrary to the laws and 
that his claim is fraudulent. That the defendant devotes so 
much of his argument to the question of fact—where he is 
weakest—only proves that it is unavoidable. 

Far from foreshadowing a separate hearing on the merits, 
this speech again, like the previous examples, closes with a 
glance at further recourse, beyond the proceedings at hand. 

 
51 Wolff, Paragraphe 24, against “Spekulationen über den objectiven 

Wahrheitsgehalt ihrer Darlegungen …; solche sind ja ohnehin fast aus-
nahmlos müßige Spielerei und juristisch zumeist ohne jeden Belang.”  

52 Paragraphe 32–33: “In this predicament, if there was any remaining 
hope for this defendant, it was this, that the court would simply vote against 
admitting the case to trial on the merits, on this account, that the plaintiffs 
could not present the document to disprove the τεκϰµήρϱια by which de-
fendant sought to render improbable the assumption of surety on his part 
(§§23–29). Perhaps he could still count on the judges to maintain a strict 
standard of proof in the paragraphê [requiring the document to prove the 
contract] where it was foremost at issue whether a contract actually existed. 
In euthydikia, … the only question in dispute would be whether that ob-
ligation came due; [in that case] there was the danger that the court would 
take a more flexible attitude toward undocumented proofs which defendant 
could not contradict with other evidence.” 


 EDWIN CARAWAN 281 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
NN speaks as though the normal expectation would be for him 
to sue the witnesses who support Apaturius: if he loses the case 
at hand, he can bring a ψευδοµαρϱτυρϱίων δίκϰη to recover his 
loss. But in that case, he protests, what basis will either side 
have to argue from, since the document itself is missing? “If he 
says it is lost, where shall I find proof that I am victim of false 
testimony?” Conversely, if the defendant had been custodian of 
the document, Apaturius would surely claim that he had gotten 
rid of it: “So why doesn’t Apaturius sue Aristocles, the one who 
took custody of the covenants and failed to produce them … 
instead of suing me and producing him as a witness?” (37–38).  
The defendant’s only recourse, if he loses the hearing at hand, 
would be to do just that, to sue Aristocles. If there were to be 
further arguments in a hearing on the merits, we might expect 
him to acknowledge it. Instead he ends with the same words he 
would use in any ordinary trial: εἴρϱηταί µοι τὰ δίκϰαια, ὅσα 
ἐδυνάµην. ὑµεῖς οὖν κϰατὰ τοὺς νόµους γιγνώσκϰετε τὰ δίκϰαια. 

In the suit of Demon against Zenothemis, Dem. 32, once 
again the argument turns upon missing evidence: the agent 
who was directly responsible for the transaction and for the loss 
would be the crucial witness, but he is nowhere to be found. 
Again, the statutory basis for the paragraphê is the law for the 
maritime courts, and this defendant invokes the same text of 
law cited against Apaturius. In this case, however, the special 
plea is grounded in the contention that “there is no contract” 
between the plaintiff and defendant.53 In Wolff’s view, that ob-
jection amounts to a procedural ploy: if an action for damage 
or ejectment were brought in ordinary court, Demon would 
have no grounds to bar the suit simply because “there is no 
contract,” and that plea in bar is dubious here. 

The background to the case is, briefly, as follows. Protus had 
borrowed from Demon (D), in order to purchase grain at Syr-
acuse and ship it back to Athens for sale. As D tells it, once the 

 
53 This is also the basis for the paragraphê in Dem. 35; on the contract in 

this case see Carawan, GRBS 46 (2006) 355–357. 


282 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
grain was loaded in Syracuse, Zenothemis (Z) conspired with 
the shipowner Hegestratus to raise further loans by pretending 
that the cargo on board was theirs (whereas Z had no legiti-
mate claim to it). Now, the borrowers owe nothing if the ship 
goes down with its cargo. So, after two days at sea, Hegestratus 
tried to scuttle his ship; he was discovered, tried to escape, but 
was drowned. The ship managed to make port in Cephallenia, 
and there Z came forward with a claim upon the cargo, urging 
that the ship should proceed to Marseille, his own and He-
gestratus’ home port. But the officials in Cephallenia ordered 
the ship to proceed to Athens.  

In Athens Protus was barred from taking possession of the 
grain by Z, who thus asserted his claim to it. Protus would then 
have recourse to “ejectment”—to take possession of the cargo. 
But Protus was reluctant to use force, and Z refused to give up 
the goods voluntarily. It was only when D arrived on the scene 
that Z agreed to go peacefully, on condition that D would be 
responsible for the ejectment. Z then brought legal action 
against Protus and against D, the latter by suit for ejectment, 
dike exoulês.54 The case against Protus was quickly resolved in 
favor of Z: as Protus did not appear before the arbitrator, the 
case went by default. Meanwhile Z proceeded against D for the 
ejectment, arguing that he had been wrongfully deprived of 
goods that he had a legitimate claim to.  

Now in framing the issue Demon explains the paragraphê as 
follows (1–2):  

The laws [for the maritime court] provide for lawsuits involving 
shipowners and merchants in obligations for shipment to and from 
Athens, and where there are written agreements;55 but if anyone pros-
ecute contrary to these provisions, the suit is not actionable (µὴ 
εἰσαγώγιµον εἶναι τὴν δίκϰην). Between me and this fellow 

 
54 On the dike exoulês see Isager, Aspects 144–147. 
55 This and similar cases point to a restrictive reading: only obligations 

based on written agreements are actionable in this court (see n.48 above); cf. 
Isager, Aspects 151–152; D. M. MacDowell, Demosthenes the Orator (Oxford 
2009) 275 with n.60, on this case. 


 EDWIN CARAWAN 283 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
Zenothemis there is neither obligation nor written agreement, as 
he himself admits in his complaint; for he says that he loaned 
money to Hegestratus, the shipowner, and when the latter was 
lost at sea, we deprived him (Z) of his cargo. Such is the com-
plaint. From the very same argument you will learn that the 
claim is not actionable and you will see that this fellow has in-
vented the whole sorry scheme.56  

The procedural issue amounts to this: Z has a claim based 
upon a maritime contract and he therefore asserts his right to 
bring suit in the maritime court; D insists that the maritime 
suits are properly reserved for obligations based on a contract 
between the two parties, not upon some incidental agreement with 
a third party. And that is not merely a technicality; for he ar-
gues at length that Z’s claim to the goods in question is entirely 
fraudulent—he borrowed money on goods that were not his. 
Wolff discounts that whole argument: once again, the paragraphê 
is a desperate last resort. Were the case to go to trial on the 
main issue, whether D had wrongly ejected Z, the plaintiff 
would surely prevail.57 But where is the contract linking D to 
Z?  

Wolff's answer is intriguing—the ejectment itself creates a 
viable obligation:58  

 
56 Dem. 32.1–2: οἱ νόµοι κϰελεύουσιν, ὦ ἄνδρϱες δικϰασταί, τὰς δίκϰας εἶναι 

τοῖς ναυκϰλήρϱοις κϰαὶ τοῖς ἐµπόρϱοις τῶν Ἀθήναζε κϰαὶ τῶν Ἀθήνηθεν συµ-
βολαίων κϰαὶ περϱὶ ὧν ἂν ὦσι συγγρϱαφαί· ἂν δέ τις παρϱὰ ταῦτα δικϰάζηται, 
µὴ εἰσαγώγιµον εἶναι τὴν δίκϰην. τουτῳὶ τοίνυν Ζηνοθέµιδι πρϱὸς µὲν ἐµὲ 
ὅτι οὐδὲν ἦν συµβόλαιον οὐδὲ συγγρϱαφή, κϰαὐτὸς ὁµολογεῖ ἐν τῷ ἐγκϰλή-
µατι· δανεῖσαι δέ φησιν Ἡγεστρϱάτῳ ναυκϰλήρϱῳ, τούτου δ’ ἀπολοµένου ἐν 
τῷ πελάγει, ἡµᾶς τὸ ναῦλον σφετερϱίσασθαι· τουτὶ τὸ ἔγκϰληµ’ ἐστίν. ἐκϰ δὴ 
τοῦ αὐτοῦ λόγου τήν τε δίκϰην οὐκϰ εἰσαγώγιµον οὖσαν µαθήσεσθε, κϰαὶ τὴν 
ὅλην ἐπιβουλὴν κϰαὶ πονηρϱίαν τουτουὶ τοῦ ἀνθρϱώπου ὄψεσθε. 

57 Cf. Paragraphe 42 with n.53, emphasizing the disparity in evidence. 
58 Paragraphe 44–45 with n.56: “prozessual ging es ja gar nicht um die (ge-

rechtfertigte oder ungerechtfertigte) Besitzergreifung des Protos, und nach 
ihm des Demon, an dem Getreide, sondern um die von letzterem vollzo-
gene ἐξαγωγή. Diese aber war gerade das formale Mittel zur Schaffung einer 
(deliktischen) Haftungsbeziehung—und diese allgemeine Bedeutung, nicht bloß 
 

284 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
In processual terms it is not a matter of taking possession of the 
grain (rightly or wrongly) on the part of Protus or Demo, but 
rather of the latter's executed ejectment. But this was just the 
formal means of creating a (delictual) obligation—and this 
general meaning, not simply “contract,” belongs to symbolaion!—
on which grounds the very party who has been wrongfully 
barred from taking possession (and this is precisely the claim of 
Zenothemis) can proceed with a dike exoulês.  

Wrongs create liabilities, indeed, and the Athenians certainly 
described such obligations as symbolaia. Moreover, Z agreed to be 
ejected by D, with the understanding that he could later bring 
suit to recover what he lost. But to suppose that this form of 
liability meets the requirement for “contract” in that jurisdic-
tion seems to me as doubtful as any of the strategies that Wolff 
imputes to the desperate defendants. Wolff supposed that the 
maritime court would normally have jurisdiction in cases of 
ejectment involving imported goods.59 But that does not quite 
square with the evidence (as most scholars now read it): the law 
for paragraphê in this jurisdiction makes it reasonably clear that 
the defendant can bar any claim that is not based on a written 
contract, syngraphai (nn.48 and 55 above). The practical effect of 
that rule would be that other kinds of claims (such as wrongful 
ejectment), without a contract to establish the claimant’s right 
to the goods, do not belong in this court. So in this case, as in 
other mercantile paragraphai, the rule defining jurisdiction is not 
merely a procedural distinction: by barring claims without a 
contract, it recognizes the defendant’s right to goods he has 
secured by contract. 
___ 
die von ‘Vertrag’, hat συµβόλαιον!—auf Grund deren derjenige, dessen 
Zugriff zu Unrecht abgewehrt worden war, und eben das nahm Zenothemis 
für sich in Anspruch, mit der δίκϰη ἐξουλῆς vorgehen konnte.” For this 
“delictual” liability (non-consensual symbolaion) see esp. Arist. Eth.Nic. 9.1.9, 
1164b.  

59 As Wolff explains, Paragraphe 45: “Insofern als nun diese der Sache 
nach allerdings eine stilisierte Form der Austragung des Streits um das Gut 
selbst war, und es sich im vorliegenden Falle um Einfuhrgut handelte, 
mochte man wohl die Zuständigkeit des Hafengerichts fu ̈r gegeben halten.” 


 EDWIN CARAWAN 285 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
If we follow Wolff’s interpretation the sole issue in the para-
graphê is the purely jurisdictional distinction:60 D has no interest 
in arguing about the facts because the preponderance of the 
evidence is on the plaintiff’s side. D would be happy to see the 
case dismissed on procedural grounds, since the maritime court 
was the only venue where Z had any case.61  

But the natural implication of this speech—as in the other 
Demosthenic speeches—is that the bearing of the contract is 
not regarded as a formality divorced from the facts. There is 
often the question of fraud. And there is always the question of 
whether the claim has been foreclosed. The formal issue is a 
question of whether there is any viable claim based on the con-
tract. But in this case and others like it, that question is bound 
up with substantive questions about the plaintiff’s right to what 
he claims: What was done to create the obligation or to undo 
it?  

The argument ends abruptly in §30 with a reference to un-
finished business.62 D accuses Z once again of conspiring with 
Protus to deprive him of his property; the proof is that he, 
Demon, will avail himself of klêteusis, whereas Z has no in-
tention of resorting to that remedy. As Wolff acknowledged, 
κϰλητεύσω probably indicates a formal commitment by D to 
proceed against Protus for “failing to appear” (lipomartyria). So 

 
60 From this perspective, treating the paragraphê as exceptio fori, Wolff dis-

misses the argument in §22 as “sophistic wordplay.” Here the speaker treats 
the term eisagôgimos/eisagein literally, as though it refers to bringing the 
disputed goods (or the culprit) into the court’s jurisdiction, and I am not so 
sure the Athenians would see it as facetious or sophistic. That literal sense of 
eisagôgimos/eisagein is also key to the commonplace in Dem. 34.43 and 35.47–
49 (see 265–267 above, 288 below). 

61 Paragraphe 43 with n.54. Wolff assumes that Z might sue in another 
court but the risk is minimal. 

62 That is, D has already summoned Protus to give testimony at the pres-
ent hearing and, of course, he fails to appear. The concluding sections, 31–
32, anticipate some accusations against Demosthenes, of no relevance to the 
main argument. 


286 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
here again, as in the case against Apaturius, the only recourse is 
to prosecute the witness; no one anticipates a separate hearing 
on the merits.  
3. Ending the dispute 

Not long after Wolff’s study appeared, MacDowell set to 
work on the second volume of The Law of Athens (1971), which 
Harrison had left unfinished. There Wolff’s model is articu-
lated at length (106–124), and MacDowell would later follow 
that model in his own handbook (1978). And with such author-
ity on his side, it is no surprise that Wolff has “demolished” 
Paoli’s hypothesis and the matter is not to open to dispute.63 
The plaintiffs present their whole case only because they will be 
denied a proper hearing if the paragraphê prevails. And, however 
much the defendants delve into the facts, they are pinning all 
their arguments on the procedural issue because they have no 
case on the merits. Thus we are to conclude that the two issues 
were clearly distinct and properly assigned to separate hear-
ings.64 But the fact that defendants would have been at a grave 
disadvantage in a hearing on the merits, or that plaintiffs had 
to present their case as though they might not get another 
chance, does not prove that there ever was a second hearing. 
That conclusion is based on the assumption (“nothing to pre-
vent our believing”) that the Athenians divided the issues and 
assigned them to separate hearings, much as later law would 
do. 

Those who rely on that premise find reassurance in some 

 
63 Harrison, Law of Athens II 111, “there is nothing … to weaken the view 

that the issue raised by the παρϱαγρϱαφή was treated quite separately from the 
main issue”; 112 “nothing to prevent our believing that the issue … was 
whether suit lay against Lakritos”; 116, “nothing to suggest that the prin-
cipal aim … was to persuade the jury on the merits …” 

64 Law of Athens II 119: “We can say with some confidence that argument 
and voting on a παρϱαγρϱαφή were quite distinct from argument and voting 
on the issue of substance and that when the παρϱαγρϱαφή was rejected a new 
hearing, quite possibly though not necessarily before a different jury, was 
opened.” 


 EDWIN CARAWAN 287 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
rather dubious evidence. There is of course the testimony of 
Pollux and the later lexica. But that tradition draws on rhe-
torical hypotheses rather than speeches-for-trial or any credible 
record of the law: the standard examples are cases rejected as 
not suited for eisangelia but for graphê paranomôn, or not for public 
prosecution but private suit, or not (for murder) before the 
Areopagus but (for manslaughter) before the Palladium court.65 
These instances have nothing to do with the paragraphê pro-
cedure at Athens in the fourth century B.C.; they illustrate issue 
theory.  

But there is an exceptio fori in the plea of Pancleon addressed 
by Lysias 23, and the testimony in Pollux has encouraged 
scholars to suppose that Pancleon’s plea is indeed a paragraphê. 
So Wilamowitz assumed though, by his reading, the plaintiff 
speaks first. It is fair to say, the order of speakers is not quite 
clear.66 But there is no doubt about the disparity: this argument 
bears no similarity to the other paragraphê speeches. As we have 
seen, the other paragraphai never rely on such technicalities; 
when such objections are raised, they always come with a dis-
claimer (nn.45–46 above). The most striking difference is 
simply that Against Pancleon makes no argument at all on the 
merits—indeed, we have no clear indication what the plaintiff’s 
claim happens to be.67 Scholars have seized on that disparity as 
proof that the hearing at hand would only decide the question 
of jurisdiction, but it seems to me at least as likely that in this 

 
65 Poll. 8.57: οἷον οὐκϰ εἰσαγγελίας ἀλλὰ παρϱανόµων, οὐ δηµοσίᾳ ἀλλ’ 

ἰδίᾳ, ἢ ὡς οὐ παρϱὰ τούτοις κϰρϱίνεσθαι δέον, οἷον οὐκϰ ἐν Ἀρϱείῳ πάγῳ ἀλλ’ 
ἐπὶ Παλλαδίῳ. See n.13 above. 

66 Thus Isager, Aspects 124 n.5, concluding (with Wolff, Paragraphe 108–
111) “that a ‘normal’ paragraphe is at issue.” 

67 As we noted at the outset (259–261 above). It is tempting to suppose 
that the case involves enslavement for debt or damages by a freedman with 
(supposedly) metic status (on which see Todd, Shape 181, 196). In the 
competing claims it is clear that Pancleon risks losing his liberty and so it is 
likely that NN’s suit may have the same effect (thus, §5, he takes pains not to 
be accused of ὑβρϱίζειν).  


288 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
singular case the main claim required no speechwriter’s assis-
tance. In the other speeches there is always an account of the 
dispute ἐξ ἀρϱχῆς. So it seems reasonable to suppose that Pan-
cleon’s plea is simply called antigraphê because it is not a para-
graphê.68 

Among all the other speeches the only passage that seems to 
point specifically to a second stage of the proceedings comes in 
the testimony with which we began, Isocrates’ prologue ex-
plaining the procedure: the archons “introduce this issue first 
and the man who brought the special plea would speak first.”69 
By itself, the first clause, τοὺς δ’ ἄρϱχοντας περϱὶ τούτου πρϱῶτον 
εἰσάγειν, might suggest that the issues are treated separately: 
the archons introduce the first question first, and then bring on 
the sequel. But the phrasing is not unequivocal, and this clause 
should not be taken apart from what follows. Apparently the 
law does not say “bring in the paragraphê first and thereafter the 
dikê.” Isocrates describes the issue elliptically: the archon is to 
“bring in” something “about this (dispute).”70 What is it that he 
brings in? 

The Athenian jury probably understood the phrase περϱὶ 
τούτου … εἰσάγειν not as an abstraction but in practical terms: 
the archon will introduce the issue that the defendant has 
raised by having his formal plea, his antômosia, read out to the 
court. To put it another way, he introduces the issue by “bring-
ing in” the defendant who raised it.71 That is not to say that the 
 

68 Pollux, in fact, in the next entry after paragraphê, notes the overlap and 
the differences (8.58, ἀντιγρϱαφὴ δέ, ὅταν τις κϰρϱινόµενος ἀντικϰατηγορϱῇ). 

69 As for the passage where the plaintiff calls defendant to answer, see 
n.29 above.  

70 Cf. the law cited in Dem. 24.54 (ne bis in idem) µὴ εἰσάγειν περϱὶ τούτων 
εἰς τὸ δικϰαστήρϱιον µηδ’ ἐπιψηφίζειν τῶν ἀρϱχόντων µηδένα; 35.51: µηδὲ 
ἀρϱχὴ εἰσαγέτω περϱὶ τούτου µηδεµία (money lost in transporting grain else-
where). 

71 E.g. Isoc. Antidosis 287 (no one has “brought in” those who encourage 
the drunken youth); Lys. 13. 28 (“bring in” defendants to the council under 
the Thirty); cf. the law in Dem. 21.47, where defendants are omitted but 
naturally implied (οἱ δὲ θεσµοθέται εἰσαγόντων εἰς τὴν ἡλιαίαν). 


 EDWIN CARAWAN 289 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
Athenians could not conceptualize the “issue” as an abstract 
question or could not speak of “introducing a dispute,” ἀγῶνα 
εἰσάγειν. But, without such language, the ordinary implication 
seems to be that the archon literally “brings in” the litigants 
and that introduction is represented in the statements that he 
has read for them.  

So, in practical terms, the likely meaning of Isocrates’ de-
scription is simply that the archon has the paragraphê read 
before the plaint. That presentation to the court comes first (of 
all), πρϱῶτον. And then the defendant speaks first or, strictly, prior 
to the plaintiff (λέγειν δὲ πρϱότερϱον τὸν παρϱαγρϱαψάµενον). This 
second rule uses the more precise way of setting one event 
before another, because the reverse order is the surprise that 
requires an explanation. If there were separate proceedings for 
the main issue, and a second decision for the jury to make after 
the paragraphê, that would also be novel and perhaps confusing, 
and we might expect some guidance on that new protocol: 
πρϱῶτον µὲν περϱὶ τούτου εἰσάγειν … ἔπειτα δὲ τὴν δίκϰην. In-
stead Isocrates seems to suppose that the jury would normally 
expect a plea of this sort, that the plaintiff is abusing the pro-
cedure, to be simply part of the defendant’s argument (as in 
Antiphon 5). Now it is presented at the outset and that reversal 
alters the dispute. But there is nothing to suggest that the jury 
are to decide the defendant’s objection first and only then proceed to 
plaintiff’s case, though that would be the crucial departure 
from the norm. If that were the order of business, surely the 
jurors should be advised that they have only to decide on the 
procedural issue at present and may reserve judgment on the 
merits until later. So, in my view, “introduce this issue first” 
simply refers to the fact that the defendant is introduced first, 
with the reading of his paragraphê, and he will begin the debate.  

After all, it takes two sides to frame an issue, and it looks as 
though the plaintiff’s side is simply his original claim against 
the defendant. If the jury were to decide the procedural ques-
tion apart from the main issue, we might expect the plaintiff to 
enter a formal reply, perhaps even a separate antômosia on that 
question: if the paragrapsamenos claims that the suit is barred by a 


290 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
settlement with full release (for instance), the plaintiff would re-
spond specifically to that challenge—presumably that the settle-
ment was invalid. The defendants sometimes begin by referring 
to the wording of the paragraphê, which the jury has just heard, 
but there is no reference to any formal response to the para-
graphê, even in the plaintiffs’ speeches (Dem. 34–35). We can 
reconstruct their positions from the arguments, but there seems 
to be no official formulation on that separate question.72 Thus, 
in the first instance (Isoc. 18), the defendant proceeds as though 
he does not know what Callimachus will say specifically in re-
sponse to his plea.73 The only answer to the paragraphê is the 
plaint. 

If we set aside the usual assumption and simply form our 
judgment from the speeches themselves, there is nothing to 
suggest that the jury must separate the two questions and de-
cide one before the other. On the contrary, the events that 
created the obligation and the course that the plaintiff has 
taken to recover what is owed to him are not easily divisible. 
Therefore, to the Athenians it seemed reasonable and sufficient 
for one verdict to answer both questions, because the paragraphê 

 
72 See esp. Dem. 34.3–5, where the plaintiff does give a fairly succinct 

reply; 35.4–5, where the plaintiff largely discounts the paragraphê and pleads, 
“If I convict him of wronging us, the lenders—and (doing wrong to) you no 
less—aid us with justice,” βοηθεῖτε ἡµῖν τὰ δίκϰαια. G. M. Calhoun, 
“Athenian Magistrates and Special Pleas,” CP 14 (1919) 338–350, supposed 
that that there must have been some formal reply to the paragraphê in the 
preliminaries, but he acknowledged that there is no indication of it in the 
speeches; in fact, Dem. 37.22–30 suggests that “the original complaint and 
the παρϱαγρϱαφή seem to constitute the pleadings” (345 n.5). 

73 Isoc. 18.7 and 13, show that C has not answered the paragraphê in the 
statement that has just been read to the court: NN says, “perhaps” C will 
claim that NN was the instigator of the confiscation (in response to the plea 
that NN is shielded by the covenant on informants); and “I learn that” C 
intends to deny the arbitration. The complaint was read in its original form, 
specifying the damages (§33). Similarly in Dem. 38.1–2 it seems clear that 
the plaintiff’s statement is his original claim for damages in the specific 
amount, ἐπὶ τῇ δίκϰῃ τίµηµ᾽᾿ ἀκϰηκϰόατε.  


 EDWIN CARAWAN 291 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
is a complete defense and the plaintiff will have made the best 
of his case, in a dispute that cannot be parsed without preju-
dice. 

For in these trials the issue is essentially whether the plain-
tiff’s claim is backed by binding agreement or foreclosed by it. 
The defendant gains the advantage of speaking first and fram-
ing the debate to his advantge, as he argues that any claim is 
canceled. If he prevails he wins the epôbelia and the plaintiff will 
pay for his litigiousness. But the plaintiff has the last word and, 
to make the most of it, he introduces all his evidence and 
follows all the implications. The jurors will vote once, for one 
litigant or the other—they do not decide on a formality. If the 
paragraphê is rejected, the plaintiff prevails and he earns the 
epôbelia in addition to his damages. That appears to be the 
working rationale when Archinus introduced the procedure, 
and we find the same principle at work in the last of the sur-
viving speeches.  

When the procedure was brand new, Isocrates argued that 
Callimachus’ claim was strictly barred by the covenants of 
reconciliation. Among those articles of agreement was the rule 
that suits and settlements be binding. That principle was re-
stated in subsequent legislation: “Whatever terms the parties 
agree to shall be final.” Thus he argues, a fortiori, when private 
agreements are enforced by public authority (τὰς µὲν ἰδίας 
ὁµολογίας δηµοσίᾳ κϰυρϱίας ἀναγκϰάζετ’ εἶναι), it is all the more 
outrageous for Callimachus to violate the city’s covenants to 
serve his own private interest (18.24–26). The Athenians swore 
to those covenants as part of a binding transaction, with stip-
ulations to foreclose any further dispute: returnees may reclaim 
much of their property, and whatever settlements they devise 
shall be final. Now, as Isocrates puts it, they are asked to violate 
their pledge in the interest of predatory litigators such as Cal-
limachus. So the jury must recognize that they are casting a 
verdict on the very viability of covenants (synthêkai) as the main-
stay of commerce and civil society (28): 

… most of our way of life, for Greeks and non-Greeks alike, 
happens through covenants. Putting our trust in them we visit 


292 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
one another and transport whatever goods we happen to need; 
with [covenants] we complete our transactions among ourselves; 
and we settle our differences, both private quarrels and wars 
that involve the whole community. This one common practice 
all mankind have always applied. So it is fitting for all to come to 
their aid, and especially fitting for you. 

In this passage the principle defended by the new procedure is 
the package of rules that the parties had agreed to. The sanctity 
of covenant is represented on two levels: it is not only the 
Amnesty at large that demands compliance, but also private 
agreements that are concluded with binding formality. For the 
Reconciliation Agreement evidently embraced the rule that 
legal judgments and arbitrated settlements (concluded under 
the democracy) shall be final, and that rule was then enacted 
into law.74 The law for the new paragraphê procedure seems to 
have come on the heels of that enactment, perhaps even as 
corollary to it. 

The later laws granting paragraphai in various venues ex-
tended the reach of that remedy, but the later speeches all 
embrace the same principle in one way or another. In the con-
tract cases, the dispute is about binding agreement at the most 
basic level. The two defendants, in Dem. 32 and 33, argue that 
they are not subject to any obligation of this sort. The essence 
of that rule is that proper contracts dispose of any further 
dispute (33.35–36): 

The crux of the matter is this: Apaturius will not even try to 
claim that he has a contract with me. When he lies and says that 
I was listed as surety in the contract with Parmeno, demand (to 
see) the contract. And then confront him on this ground, that all 
men, when they make contracts with one another, seal the docu-
ment and put it in the custody of those they trust, for this reason 
—so that if they have any dispute, they have recourse to the text 
and on that basis they can put the matter in dispute to the proof. 

 
74 With Isoc. 18.24 (ἰδίας ὁµολογίας δηµοσίᾳ κϰυρϱίας) compare Andoc. 

1.87–88, among reforms inspired by the Amnesty; Dem. 24.54, canceling 
decisions under the Thirty; cf. Carawan, GRBS 46 (2006) 368–372.  


 EDWIN CARAWAN 293 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
The two plaintiffs, in Dem. 34 and 35, argue on the other side, 
that contracts must be enforced against defendants who have 
done all they can to evade their obligations. Both parties argue 
the merits because questions of fact and fraud are inseparable 
from the question of whether the claim has any basis in binding 
agreement. The plaintiff against Lacritus frames the issue as 
follows (35.26–27): 

What is agreed by both parties, in covenants for which a mari-
time contract is secured, everyone considers final (τέλος ἔχειν); 
one has to apply what is written. That they have done nothing 
according to the contract but from the very beginning com-
mitted fraud and plotted to do wrong, is proven by the witnesses 
and by their own claims. 

But the clearest instances come from the last three speeches in 
the set, Dem. 36–38: in each case prior settlements, solemnized 
with release and quittance, bar any further dispute; that basis is 
set forth in the prologues (as we saw, 274 ff. above). And the 
principle of finalty is thoroughly developed in defendants’ argu-
ments. Thus in the epilogues to the last two speeches (37.58–59 
= 38.21–22) we find it framed as a commonplace: even in mat-
ters of bloodshed—the most unquenchable grievances—once 
the parties have reconciled and given release to the killer, there 
must be a boundary barring any further recrimination. It is the 
same principle embodied in the closing to the Amnesty (cf. 
Dem. 40.46). 

Again the first case is perhaps the best illustration (Isoc. 18). 
In their original settlement NN conceded his share in the loss, 
as he had been present when Patrocles took the money; Cal-
limachus settled for that concession and gave a release from 
further claims. That should have ended the matter. But it is 
clear from NN’s presentation that C has changed his con-
struction of the events: presumably he claims that NN was the 
main actor and now fully liable; on this issue, C contends, 
“there was no diaita.” That at least is a plausible interpretation 
of what NN says: C was able to convince the archon to reopen 
the case by arguing that this is a new complaint of a more 
serious wrong, one that was (arguably) exempt from the agree-


294 PARAGRAPHÊ AND THE MERITS 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
ment and from the covenants of reconciliation. 
Evidently, by claiming that NN is not a mere accomplice, C 

hopes to evade the covenant shielding “informants and de-
nouncers.” There were no definitions in the law, but, to make a 
good case, C probably needed to show that NN was the prime 
mover, acting from his own motives, to his own advantage. It 
looks as though the evidence on that is doubtful or C’s con-
struction is precarious, based on NN’s ties to the discredited 
regime. That is why our defendant insists upon it:75  

If he mentions what happened under the oligarchy, demand that 
he not accuse those whom no one will defend. Demand instead 
that he show that I—the person you must vote on—took the 
money. Demand not that he show that he has suffered terribly, 
but that he substantiate that I caused it, (as I am) the one from 
whom he is demanding to recover his losses. 

Whereas Paoli treated this passage as proof that the main claim 
will be decided in the hearing at hand, Wolff discounted it as a 
rhetorical appeal to a general amnesty. But the issue is more 
clearly defined: the defendant insists upon this point precisely 
because there was a covenant specifically barring any claim of 
mere complicity in the crime; and the alternative, the charge 
that NN (or anyone) “caused it,” may be very difficult to prove. 
C will argue that NN was the main actor, who set the events in 
motion, to his own advantage. But that picture involves a 
complicated reckoning that NN asks the jury to reduce to its 
simplest terms. And it is all the more burdensome for the 
plaintiff, if the jurors weigh the prior agreement that Callim-
achus made with this same defendant: How can Callimachus 
claim such a sum for a loss he has already settled in arbitration, 
pledging “no further dispute”? 

Awkward as it is, the jury’s decision does not involve sophis-
ticated distinctions of fact and form. It is the sort of problem 
that neighbors and business partners wrestled with all the time, 
settling their own disputes and defining their obligations. There 

 
75 Isoc. 18.40, again following Mirhady’s translation.  


 EDWIN CARAWAN 295 
 

————— 
Greek, Roman, and Byzantine Studies 51 (2011) 254–295 

 
was no elite judiciary to parse these complexities, and the ar-
chons had no competence to dispose of them summarily. These 
were cases for a jury representing the community to decide, not 
on a formality but for one litigant or the other. 
 
March, 2011        Modern and Classical Languages 
         Missouri State Univ. 
         Springfield, MO  65897 

         ECarawan@missouristate.edu