Legal Reasoning and Legal Theory Revisited Author(s): Fernando
Atria Source: Law and Philosophy, Vol. 18, No. 5, Laws, Facts, and
Values (Sep., 1999), pp. 537-577 Published by: Springer Stable URL:
http://www.jstor.org/stable/3505144 Accessed: 31/08/2010 11:37Your
use of the JSTOR archive indicates your acceptance of JSTOR's Terms
and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's
Terms and Conditions of Use provides, in part, that unless you have
obtained prior permission, you may not download an entire issue of
a journal or multiple copies of articles, and you may use content
in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this
work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=springer.
Each copy of any part of a JSTOR transmission must contain the same
copyright notice that appears on the screen or printed page of such
transmission. JSTOR is a not-for-profit service that helps
scholars, researchers, and students discover, use, and build upon a
wide range of content in a trusted digital archive. We use
information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected].
Springer is collaborating with JSTOR to digitize, preserve and
extend access to Law and Philosophy.
http://www.jstor.org
FERNANDOATRIA*
LEGAL REASONING AND LEGAL THEORY REVISITED (AcceptedMay 20,
1999)ABSTRACT. This article deals with the relationbetween a theory
of law and a theory of legal reasoning.Startingfrom a close
readingof ChapterVII of H. L. A. Hart's The Conceptof Law, it
claims that a theory of law like Hart'srequires a particular theory
of legal reasoning,or at least a theory of legal reasoningwith some
particularcharacteristics.It then goes on to say that any theory of
legal is reasoningthatsatisfiesthose requirements highly
implausible,andtries to show thatthis is the reason why not only
Hart,but also writerslike Neil MacCormick and JosephRaz have failed
to offer a theoryof legal reasoningthat is compatible with legal
positivism as a theory of law. They have faced a choice between an
explanationof legal reasoning that is incompatiblewith the core of
legal positivism or else strangely sceptical, insofar as it severs
the link between general rules and particular decisions thatpurport
apply them. to KEY WORDS: application of law, appropriateness,clear
cases, deductive reasoning,legal positivism, legal reasoning,open
texture
When H. L. A. Hartwrote The Conceptof Law, legal reasoningas
such was not in the philosophical agenda. Consequently,he later
acknowledgedthatin The Conceptof Law he had "saidfar too little
aboutthe topic of [...] legal reasoning"(1994, p. 259). This aspect
of Hart's book was soon subject to criticism because it appeared to
some (most notably, Dworkin 1967) that the implications of Hart's
theory for legal reasoning were clearly at odds with what lawyers
andjudges saw themselves as doing. What was needed, in to
consequence,was a "companion" The Conceptof Law, an examination of
the way in which a powerful explanationof the nature of law such as
Hart's could furtherthe understanding only of not what the law is,
but also of how the law works, or, better, how* Licenciado en
Derecho (Universidadde Chile, 1994); Ph.D. (Universityof
Edinburgh,1999); Assistant Professor of Law, Universidadde Talca
(Chile). I am indebted to Zenon Bankowski, Neil MacCormick,Kevin
Walton, Claudio Michelon and Burkhard Schafer for
thoughtfulcriticism of previous versions of this article.They have,
of course, no responsibilityfor the mistakesthatremain.LA Law
andPhilosophy 18: 537-577, 1999.
9
)? 1999 KluwerAcademicPublishers. Printed in the
Netherlands.
538
ATRIA FERNANDO
people work with the law: a theory of the applicationof the law
(i.e. legal reasoning). We are now told that Legal Reasoning and
Legal Theorywas supposedto be such a companion(MacCormick 1994, p.
xiv). Before consideringthat book, let me explain where the tension
between legal reasoningand legal theorylies, and to do so we shall
considerH. L. A. Hart's"opentexture"thesis.HART ON OPEN TEXTURE
For our purposes, a full exposition of Hart's open texturethesis
is not needed here: suffice it to say thathe triedto strikea middle
way between whathe called "ruleformalism" "rulescepticism",and and
that to do this he borrowedfrom F. Waismann(Waismann1951) the idea
of open texture.According to the text-book exposition of this
thesis, the argumentwas that, since meaning is use, conceptwords
cannot have any meaning whatsoever without there being clear
instances to which they apply. To be able to recognise those
examples as instances of the relevantconcept-word(s)is to know the
meaning of the latter. By the same token, however, in many
instancesthe applicationof those concept-wordsto some event will
not be completely obvious, and disagreementbetween competent users
will arise. In these circumstances,failure to use the relevant
concept-wordto refer to those events is not evidence of ignorance
of theirmeaning (as failurein the clear cases is), since these
events are said to be in the penumbraof meaning of the
relevantconceptword(s), where different opinions might exist
between competent users as to whether or not a
particularconcept-wordapplies. To make an often-quotedpassage even
more often quoted,If we are to communicatewith each otherat all,
and if, as in the most elementary form of law, we are to express
our intentionsthat a certaintype of behaviourbe regulatedby rules,
then the generalwords we use [... ] must have some standard
instancein which no doubtsarefelt about[their]application.
Theremustbe a core of settled meaning,but there will be, as well, a
penumbraof debatablecases in which wordsareneitherobviously
applicablenor obviouslyruledout (Hart1958, p. 63).
So understood,Hart's is a thesis about the limits of
certaintythat general classificatoryterms can have in
naturallanguages:"[open textureis] a generalfeature of
humanlanguage;uncertaintyat the
LEGALREASONINGAND LEGALTHEORYREVISITED
539
borderlineis the price to be paid for the use of general
classifying terms in any form of communicationconcerning matters of
fact" (1994, p. 128; my italics). It is an inescapable feature of
natural languagesas we know them, andhence is partof the
humanpredicament:if we areto communicatewith each otherusing
natural(rather than artificial) languages, then it is pointless to
strive to achieve complete certainty: there is nothing we can do to
exclude open texture,at least insofaras we also want to use
generalclassificatory terms:my view was (and is) that the use of
any language containingempiricalclassificatorygeneral terms will,
in applyingthem, meet with borderlinecases calling for fresh
regulation.This is the featureof languageI called 'open texture'"
(Hart, quotedby Bix 1993, p. 24).
On this first reading of it, the open texture thesis is one
about language,and only derivativelyaboutthe law. "Opentexture"is
not a featureof law but, as Hart explicitly says in the quoted
passage, one of naturallanguages. Needless to say, since (or: only
because) legal rules are expressed in naturallanguages, the open
texture of the latter communicates,so to speak, to the former.Thus
it is not surprisingat all to hear from Hart that, for example,
"whichever device, precedent or legislation, is chosen for the
communication of standards of behaviour, these, however smoothly
they work over the great mass of ordinarycases, will, at some point
where their applicationis in question, prove indeterminate" (Hart
1994, pp. 127-128). Immediatelyafterpresentingthe notionof open
texture,andin an effortto cheer the readerup,
Hartexplainsthatuncertainty apparent at the borderline
certainlynothingto be afraidof. But in the course is of this
consolationthe natureof the open-texturethesis switches: it becomes
a thesis no longer aboutone of the inescapablefeaturesof
naturallanguages as we know them, but about the convenienceof
having open-textured(i.e. not completely certain and predictable)
rules. It ceases to be a featureof languageto become one of the
law. Of course, there is no reason why you cannot argue that X is
the case and then go on to arguethat X is also desirable,which is
the usual way in which the relevant passages on The Concept of Law
seem to have been read. But Hart did somethingmore: when his
arguingaboutthe desirabilityof open texture,and contradicting
540
FERNANDOATRIA
statementsquoted above (and many others) Hart conceded that it
is possible, for us now and here, to eliminate the uncertaintyat
the borderline,i.e. "to freeze the meaning of the rule so that its
general terms must have the same meaningin every case where its
applicationis in question"(1994, p. 129). He even explainedto us
how:To secure this we may fasten on certain features present in the
plain case and insist that these are both necessary and sufficient
to bring anything which has them within the scope of the rule,
whateverotherfeaturesit may have or lack, and whatevermay be the
social consequencesof applyingthe rule in this way (1994, p.
129).
And if we were to follow his advice,we shall indeed succeed in
settling in advance,but also in the dark,issues which can only
reasonablybe settled when they arise and are identified(1994, p.
130; my italics).
If we can indeed succeed in settling in advancethe outer limits
of the law, it follows that the explanationof the fact that these
limits are uncertainmust be in the reasons why it is not
convenientfor us to do so, i.e. in the reasons why these cases "can
only reasonably be settled when they arise and are identified".In
other words, if we can eliminatethe uncertaintyat the
borderline,then it is simply wrong to say thatthe reasonwhy the law
is uncertainis because the uncertaintyat the borderlinecannot be
eliminated;the reason why the law is uncertainin hard cases is not
some inescapablefeature of general classificatoryterms in
naturallanguages, but the very differentone that it is
unreasonableto try to settle "in advance,but also in the
dark"issues we cannotyet identify. Hartexplains that Following this
second line of argumentation, he is dealing not with a limitationon
the levels of certaintyimposed on human beings by the language they
(we) happen to have, but with the very differentissue of strikinga
rightbalancebetween two competingsocial needs, i.e.the need for
certainrules [... ] and the need to leave open, for
latersettlementby an informed,official choice, issues which can
only be properlyappreciatedand settled when they arise in a
concretecase (1994, p. 130).
And furthermore, this tension is one that "infact, all systems,
in different ways" solve reaching some kind of compromise
(ibid.
LEGALREASONINGAND LEGALTHEORYREVISITED
541
Italics added). Open texturethen is not an externallimit
language imposes on the levels of certaintyhuman beings can
achieve, but the consequence of a normative decision, i.e. a
decision about how best to balance the requirementsof certainty
with those of appropriateness.1 It is importantto emphasise that,
as we have seen, what a hard of (clear) case is varies accordingto
each of these interpretations the open-texture thesis. In the
firstinterpretation, case will be hard a when the facts are such
that they do not fit naturallyand uncontroversiallyone or more of
the general classificatoryterms of the relevant rules, i.e. when it
is what could be called a "semantic" hardcase (when, e.g. the rule
forbidsyou to go into the parkwith a vehicle andyou wantto use a
toy car in it). On the second interpretation, however,the point is
not uncertaintyat the borderline.As we can "indeed succeed" in
having clear and certain rules (regardless of the features of
naturallanguages), a case will be hard because what is an issue is
not the classificationof particulars in the world, but the very
different one of whether or not this case was one of those left
"open,for latersettlementby an informed,official choice" even if it
is coveredby the semanticmeaningof the rule in question (was the
case of the veteran'swanting to use a militarytruck in a memorialin
the parksettledwhen the "no vehicles"rule was issued or was it
"left open"?).Since the only reason why it makes sense to begin to
discuss whetheror not this case is covered is that it could be
covered, in this explanationa hardcase in fact presupposesthat in
the particulars the world can be classifiedunderthe rule's general
classificatoryterms. Bearing these considerationsin mind, we go
back to the tension between law and legal reasoning, to the
challenge mentioned by MacCormick:a Hartianexplanationof legal
reasoning has to be seen to flow from, or at the very least to be
consistent with, the centralclaims Hartmade in the "mother"
theory.I hope it is not very to controversial say thatone of the
centraltenets of Hart'stheory of1 Some (or: stipulationswill be of
use here:(a) I will call "certainty" "predictthe firstof the social
needs Hartdistinguishedand (b) "appropriateness" ability") the
second; (c) I will talk of "application"when referring to the
problem of whetheror not a rule shouldbe appliedto a particular
case, and (d) of "meaning" when referringto thatof graspinga rule's
meaning.
542
FERNANDO ATRIA
law was thatat a conceptuallevel law is independentfrom
morality, that is, what the law ought to be is not the same than
what the law is.2 These two questionsare, in Hart'sview, not only
different,but logically different:it is possible to establishwhat
the law is without inquiryinto what the law ought to be; no
conclusion aboutwhat the law is follows fromarguments aboutwhatthe
law oughtto be. At the same time, Hartsaw thatany
theoreticalelucidationof the natureof law must explainwhy and how
it is possible for competentlawyers, judges and lay persons to
disagree not only about what the law in ought (morally)to be, but
also (and much more importantly this aboutwhat the law actuallyis.
Now, the explanationof the context) latterkind of
disagreementcannotbe groundedupon the existence of
disagreementaboutwhat the law ought to be, since if that were the
case the law as it is would not be conceptuallydifferentfrom the
law as it ought to be (thatis, it cannotbe the case that we
disagree about what the law is because we disagree what the law
ought to be, if these two questionsare conceptuallydifferent).Hence
we got Hart'sopen texturethesis. The importanceof this thesis is
that it performedthe role of thus supplyinga
morallyneutralexplanationof legal disagreement, allowing us to
explain disagreementaboutwhat the law is in a way thatwas not
parasiticalon disagreement aboutwhatthe law oughtto be. This was,
therefore,the explanation(at least the kindof explanation)
requiredby Hart'stheoreticalassumptions,if his theory was to have
any consistency.But Hartnoticed (or so I claimed) thatthe idea of
open texture,importantas it might be, did not explain the whole of
the fact of legal disagreement when looked at from a legal i.e.
clarificationof the meaning of words is reasoning-perspective, not
always the kind of informationthat would be useful to lawyers
andjudges and lay persons when they are discussing what the law2
There is some discussion as to the precise content of what is
sometimes
called "the separabilitythesis" (see, among others,FtiBer1996;
Coleman 1996). would not affect some versionsof This has to be kept
in mind, since my argument the thesis.
ConsiderShiner's(admittedly"crude") version:"theexistence of law is
one thing and its meritor demeritanother" (Shiner1992). I do
believe (along with most positivists, naturallawyers, and realists
of differentdenominations)that in this sense the thesis is true.I
think,however,thatI can bypassthis debatebecause in any plausible
readingthat thesis must mean, for legal positivists, that the fact
thatthe law ought to be differentis not enough to establishthatit
is different.
LEGALREASONINGAND LEGALTHEORYREVISITED
543
is in concrete cases. Hartrealised that in many of these cases
what x was discussed was not whether a particular was an instance
of a general X, but ratherwhether or not a particular(otherwise
clear and unambiguous)rule was, in a legal sense, meant to be
applied to the facts that configuredsome concrete case. Hence he
offered, in the same pages of The Conceptof Law, a second
explanationof the fact of legal disagreement,one based on the claim
that there is a built-in tension in law between (what I called)
predictabilityand appropriateness. Now, it is in my view a crucial
point that the legal theoryimplicationsof this second
explanationare at odds with the central claim of Hart'sbook
identifiedabove. In the firstexplanation,what made a case hard was
a "value-free"feature, i.e. the open texture of the relevant words.
This is why Hart was free to say that in clear cases the
applicationof rules does not require the decision maker to exercise
a "freshjudgement"(1994, p. 135). From the universe of cases courts
will have to solve from now on, some of them are (or will
eventuallybe) markedby the fact thatthey belong to the penumbraof
meaning of the relevantwords; the identification of those cases as
hardwill not imply, therefore,that evaluative ideas about what the
law ought to be will be smuggled in at the moment of
ascertainingwhat the law is. When the "mark" open of texture is
discovered the court will have reached the outer limits of the law:
it can then discuss about what the law will be after the
court'sdecision, in the light of whatthe law shouldbe, only because
there is no law on the matter.Notice that nothing guaranteesthat
this will be uncontroversial. Therecan be disagreementon whether
skateboardsand push-chairsare "core" or "penumbra" instances of the
word "vehicle".That is to say, I think Raz is correct when he says
(Raz 1985, p. 218) that it is false "thatall factual matters are
non-controversial" that "all moral propositionsare controand
versial".What is importanthere is not that accordingto the open
texture thesis the applicationof the law is non-controversial, but
thatany legal disagreementwill not be moralbutfactual (or verbal,
or conceptual)disagreement:are push-chairsand skateboards, a as
matteroffact, vehicles? The second explanation(legal disagreementas
the consequence of the tension between predictabilityand
appropriateness) does not
544
ATRIA FERNANDO
work so nicely, though it representsmore faithfully the reality
of that singles a case out as hardis legal reasoning.In it, the
"mark" an evaluativefeature:the case is (will be) markedas hardif
predictability's requirementsare overriddenby those of
appropriateness, i.e. if the solutionofferedby the rule is
inappropriate enoughfor the demandof predictabilityto be defeated
in the case. Notice that, in this view, to "discover"the "mark"
that would allow us to classify a case as clear or hardis to
exercise a "freshjudgement",as it is to answerthe questionof how
pressingthe inappropriateness a norm of ought to be for the
demandfor predictabilityto be overridden,the answerof which will
dependon the relativeimportance those values are takento have.
Fromthis standpoint questionof what the law the is cannotbe
differentiated from thatof what the law ought to be. In other
words,for the court the question "is this prama vehicle?"is linked
to the question "oughtthis pram be considereda vehicle?"
(considerthe commonjudicial way of posing this kind of problems:
"should skateboardsbe considered as vehicles for the purpose of
this law?").3 If this is correct,there is no way in which we can
say that there is a logical distinctionbetween these two
questions.To see why, it seems useful to divide Hart'sview on
hardcases up into two parts: one that contains a test about what
makes a hard case hard, and anotherthat explains what is going on
once a case is recognised as such. We have seen thattwo answerscan
be foundin TheConceptof Lawfor the firstproblem,the test thatmakesa
case hard.The answer to the second problemis that in hard cases
there is no settled law, hence the courtshave to exercise
discretion.Now I believe (though I will not argue for this here)4
that the first, non-moraltest for the firstproblem,i.e. the open
texturethesis understood a thesis about as rather thanaboutthe
law,has to be rejected,andsomething language and alongthe lines of
the tensionbetweenpredictability appropriate3 The fact that
Harthimself sometimes (e.g.: 1967, p. 106) phrasedthe question in
these terms(as one of ascertaining whethera particular is an
instanceof x a generalX for the purposeof a given law) shows thathe
failed to notice thathe was offering two explanations.If his open
texturethesis (understoodas a thesis aboutlanguage)is true,then
therearecore instancesthatarerecognisableas such regardlessof the
purposeof any law: see Schauer1991, p. 212ff. 4 See my "Gamesand
the Law: Two Models of Institution"in ARSP (July 1999).
LEGALREASONINGAND LEGALTHEORYREVISITED
545
ness must be placed there instead. If we then retain the
original second part,i.e. the claim that in hardcases courtshave
discretion, the incompatibilitybetween what we would then get and
the core of Hart'sphilosophy of law (as identifiedabove) is
evident:In this modified version, Hart'sview on hardcases would be:
(i) a case is hardwhen the applicationof the (prima facie) law is
deemed objectionable (i.e. when the primafacie solutionis such
thatthe demand for appropriateness strongerthan the demandfor
certainty);(ii) is when a case is hard,the law is unsettled,and the
courtshave discretion. In short:when the applicationof an
otherwiseclear legal rule to a case thatbelongs, so to speak, to
its core of meaningproduces an objectionableresult, it is the law
that there is no law on the subject. What the law is for the case
depends upon what the law (i.e. the balance between
predictabilityand appropriateness) ought to be for the case. When
the (primafacie) law ought to be different, itis different.lex
iniustanon est lex! Now, is this "ought"a moral "ought"?It might
appearthat the answeris obviously yes: the point is why is
predictability important, As andwhy shouldwe care
aboutappropriateness. Raz has claimed, they cannot but be moral,
since "thereis no otherjustificationfor the use of an
autonomousbody of considerationsby the courts" (Raz 1993, p. 318).
But we should be careful here. Hart is indeed careful not to talk
of these values as moral values, at least in The Conceptof Law. And
in "Positivismand the separationof law and morals"he is explicit in
denying that this is a moral ought: "we shouldrememberthatthe
baffledpoisonermay say, 'I oughtto have given her a second dose' "
(Hart 1958, p. 70). Hartalso points out that "underthe Nazi regime
men were sentencedby courtsfor criticism of the regime. Here the
choice of sentence might be guided exclusively by considerationof
what was needed to maintainthe state'styrannyeffectively"(ibid.).
So Hartbelieved that the solutionto the conflict between approcan
priatenessand predictability be based upon purelyinstrumental But
considerations. the obvious questionis, what arethese considerfor.
ations instrumental In the poisoner'scase, they are instrumental to
achieve a goal previouslyand independentlygiven, i.e. to kill the
to woman. In Nazi Germany,they "might"have been instrumental
546
ATRIA FERNANDO
the independentlygiven goal of maintainingthe state's repressive
apparatus. That the goal is previously given means, obviously, that
there cannot be an instrumental "ought" before the goal has been
specified. Therefore, when the applicationof the law is at issue,
that goal cannot be something like "to follow the law", since the
court is tryingto establishwhat the law is for the case (it would
be equivalentto saying that the goal for the poisoner is to
administer the poison, a goal that is useless as a guidance for the
problemof how muchpoison "ought"to be administered). The important
point is, how can the courtestablishwhat the goal is? Notice that
the answer is not to be found in anotherrule of the system,because
of the same reasonwhy Hartarguedthat"canonsof could not succeed in
eliminatinguncertainty(1994, interpretation" p. 126): because those
rules would also be subjectedto the problem. Still, is this a moral
"ought"?I hope we can see now that we do not need a positive answer
to this question (positive though I think the correct answer is).
Hart's answer to the rule-scepticwas to claim that the
applicationof rules to a majorityof cases does not call for
"freshjudgement",with the obvious implicationthat the
decision-makerdid not have discretionto solve those ("core") cases.
But this is an answer that works only if the first version of the
open texturethesis is accepted,since only this version does not
make the characterisation a case as clear or harddependant of upon
the very kind of judgementthat is supposedto be absentfrom one kind
of cases. In the second version, however,since the distinction
itself is basedon such a judgement,Hartis left with no answerto the
or the rule-sceptic.Howevercontroversial uncontroversial decision
to characterise case as "clear" "hard" a or it is a judgement
mightbe, for the makingof which the courthas discretion.55 We could
go furtherinto this point, though for the argumentpresentedin this
article that is quite unnecessary.Hart's claim that the "ought"in
question in instrumental achievingsome (independently mightbe an
instrumental "ought", given) goal seems to leave him ratherclose to
Dworkin:since what the goal is and how best to achieve it will
determinehow the balancebetween predictability and appropriateness
to be solved in the particular has case, whatthe law is for this
case will dependon the identificationof that goal, and the chosen
goal will have an impactupon the contentof the rules (rememberthat
we cannotknow whether
LEGALREASONINGAND LEGALTHEORYREVISITED
547
In brief, the fact that the law ought to be different sometimes
makes it be different.In order to refer to what Hart called "fresh
judgement" without giving the impression that too much
argumentativeweight is placed upon its being a moral problem,I will
use broaderexpressions, like Hart's quoted "freshjudgement",or
"evaluative judgement"and the like. The tension between legal
theory and legal reasoning is explained,at least in part,by a
differencein perspectivebetween the two: when buildinga legal
theory,what is at the centre of attention is a set of questionslike
"whatis the law?""whenare we entitledto say thata legal system is
valid (exists)?""howcan we know whether a particular is partof this
or that(or of any at all) legal system?" rule (see Raz 1980, p. If,
for a useful typology of the questions a legal theory must answer
to be a "complete"legal theory). At this level it is hardto deny
the differencebetween the law that is and the law that ought to be.
The mere fact thatmany people can sensibly think the law of
theirlandto be unjust,thatis, differentfrom whatit ought ideally to
be, shows thatthereis such a distinction. But when the focus of the
enquiryis shifted to legal reasoning, this clear difference is
upset. It is still possible to apply a law that is differentfrom
the law that ought to be, and many times judges decide one thing
while at the same time they think that a different decision ought
to have been but for the content of the applicable law.6 But we
have seen that in order to apply the law, the judge has to decide
how best to balance the values of predictabilityand in
appropriateness the instant case. The obvious fact that judges are
sometimes compelled to decide a case in a mannerthey think is not
(morally) the best shows that the law does indeed pre-empt some
substantiveissues that would otherwise be promptedby the case. But
the equally obvious fact that a law does not exclude all the
substantiveconsiderations(e.g. the considerationthat the manthe
case is clear or harduntil we have solved this tension). This seems
a different concept (see Dworkin 1986, p. 46ff). way of saying
thatthe law is an interpretive 6 Hence the italicised last at the
end of the paragraph eight paragraphs phrase back was, in a way, a
rhetoricalexcess. But it was only exaggerated,not plainly false:
sometimes laws that produce unfair or unjust results when applied
to a particularcase are not laws for that case, and that suffices
to put in question of the separabilitythesis, accordingto which
from the fact that a legal solution is morallyobjectionableit does
not follow thatit is legally mistaken.
548
FERNANDO ATRIA
who shed blood in the streetsof Bologna was a barber,and that he
was shavinga customer),even when it primafacie appearsto do so
(Thus, for example, the Bolognese statutesaid that the words had to
be taken literally,without interpretation: the case in Pufensee
dorf 1688: 5.12.8, pp. 802-803 [547]) shows that there is more to
the ascertainingof what the law is than getting the meaningright.
And it is somewhatironic that Harthimself gave such an accurate
descriptionof what this something else is, that is, the solution of
a tension between the values of predictabilityof judicial decisions
and their appropriateness the particularcase at hand. To repeat: to
what the law is for the case cannot be known before deciding how
the competitionbetween predictability appropriateness and ought to
be resolved. One could, I suppose,insist on the idea thatthis is
not a problem, and to do so one would have to arguethatan answerto
the question "what is law"? does not have any consequences for an
answer to thatof "whatis the law for this case?".If it could be
arguedthatan answerto the firstquestiondoes not imply an answerto
the second, this article'sargument would be
conceptuallymistaken.And indeed, it has been claimed that "it has
been a central presupposition[of that analytical jurisprudence]
thereis a clear distinctionbetween the philosophicalquestion,
"Whatis law?" and the lawyer's question, "Whatis the law for this
or thatmatter?" (Marmor1995, p. v). Now, there is an obvious sense
in which these are two different questions,i.e. in the same way
that the question "Whatis cancer?" is differentfrom the
question"Does this personhave cancer?".But this is not to say that
an answerto the first questiondoes not imply (at least partof) an
answerto the second, in the same way in which the answer to the
first of Marmor'squestions implies (at least part of) an answerto
the second (indeed,to say thatx has consequences for y assumes the
existence of a [more or less] "cleardistinction" betweenx andy).
One could claim, however, that the relation is not that close.
ConsiderHans Kelsen's position:This determination[of a lower-level
norm by a higher-levelnorm] however, isnever complete [... ]. Even
a meticulously detailed command must leave a number
of determinations those carryingit out. If official A
ordersofficial B to arrest to subjectC, B must use his discretionto
decide when where and how he will carry out the warrantto arrest C;
and these decisions depend upon external circum-
LEGALREASONINGAND LEGALTHEORYREVISITED
549
stances that A has not foreseen and, for the most part, cannot
foresee (Kelsen 1934, p. 78). That a judicial decision is based on
a statutemeans in truthsimply that the decision stays within the
frame the statute represents,means simply that the decision is one
of the individualnorms possible within the frame of the general
norm, not that it is the only individualnorm possible (Kelsen 1934,
p. 80. The passage remainsunalteredin the second edition of ThePure
Theoryof Law).
One could understandKelsen here as saying that the Pure Theory
of Law will never be able to answer "the lawyer's question"if that
questionis "When,where and how shouldB arrestC?".But this is not to
say thatthe answerto the firstquestionis not an answerto the
second:in Kelsen's example,a completeanswerto the firstquestion
(somethinglike "a legal system is the set of all the laws enactedby
the exercise of powers conferred,directlyor indirectly,by one basic
norm")7implies an answer to the second ("for this case, the law is
that C should be arrestedby B, though the law does not specify
precisely where, when or how"). It is not clear to me whether
Marmor was claiming that for the analyticjurisprudence two
questionswere differentin the sense that an answer to one did not
imply an answer to the second, or only thatthey were
different,withoutany furtherclaim. In the latter sense, he is
surelyrightbut it would not be an objectionto my main argumentin
this article;in the former,it would indeed be an objection but (I
would claim) it would not be true as regards"analytic nor
jurisprudence" would it be correctin its own terms.
DEDUCTIVE REASONING, CLEAR CASES AND LEGAL ARGUMENTATION
The challenge for a complete Hartian (-like) theory of law (that
is, a Hartian(-like) theory of law and legal reasoning) is, then,
to harmonisethese two perspectives,that of legal reasoning and that
of legal theory.I want to considernow in some detail what is
probably the most sophisticatedattemptto meet this challenge, i.e.
Neil MacCormick's Legal Reasoningand Legal Theory.7 This is Raz's
version of what he calls Kelsen's "criterionof identity"(see Raz
1980, p. 95; and Kelsen 1934, p. 59ff).
550
FERNANDO ATRIA
That MacCormick'sis an attemptto meet this challenge is clear
from the new foreword of the paperbackedition, where he says that
"the analyticalpositivist approachto legal theory espoused by
Hartis open to challenge, and has been challenged, for an alleged
inability to give a satisfactory account of legal reasoning,
espeThis cially reasoning-in-adjudication. book took up that
challenge" I (MacCormick1994, p. xiv). In particular, take his
argumentation the role of deductive reasoning in law as
constituting concerning the best available analysis of clear cases
in the traditionof legal positivism. This is the reason why, before
consideringMacCormick'sargument,it is necessaryto addressthe issue
of syllogistic (or deductive) reasoningin HerbertHart'stheory of
law. Harthimself sometimes showed little sympathy for the idea that
legal decisions can be he reachedin a deductivemanner:
arguedthat"logicis silent on how to classify particulars" (Hart
1958, p. 67). Commentingupon this and relatedpassages from
Hart'swork,Marmorclaims thatnothing could be fartherfrom Hart'smind
than the idea of the application of a rule to a clear case being a
matter of logic or analyticity.8 Defending Hart, Marmorhas claimed
that "it is easily discernible that whatever it is that connects a
rule to its applicationcannot consist of logic [... ]"9 and he then
argues,as Hart put it, "logic is silent on how to classify
particulars" it is precisely but this classificationto which his
distinctionbetween core and penumbrapertains. In otherwords,we
mustkeep separatewhatmightbe called "rule-rule" "ruleand
world"relations;logic [... ]10 pertain[s]only to the former,not to
the later kind of relation(Marmor 1994, p. 128).
And he concludes by saying that "neitherHartnor any other legal
positivist must subscribeto the view that the applicationof legal
rules is a matterof logical inference"(ibid., at 128).8 As we shall
see, MacCormickdoes not mention the idea that the judicial
syllogism is analytic.Of course it is, but this is not to say that
it is "analyticity" (or "logic")whatconnects a rule to its
application.I will arguethatwhatconnects a rule to its
applicationis logic plus the distinctionbetween core and penumbra.
For this reason,I will follow MacCormickin not discussingthis at
all in termsof analyticity. 9 omitted. Following the previousnote,
"andanalyticity" 10 "And omitted. analyticity"
LEGALREASONINGAND LEGALTHEORYREVISITED
551
Marmoris rightwhen he claims thatthe distinctionbetween core and
penumbrais not a matterof logic, but let us ask the question:
importantfor Hart?"And the answer "why is the core/penumbra is:
because, in addition to the existence of a core and penumbra of
meaning for most (all) concepts, Hart claimed (at least in the of
traditionalinterpretation the open texturethesis) that a state of
affairsconstitutesa clear legal case when in some of its
descriptions it is encompassedby the core meaningof some
applicablerule, and hardotherwise.It is with this furtherclaim
thata space for logic and deductivereasoningappears:to put it in
Marmor'sterms, once the of relationrule-worldhas been settled, once
the particulars the case have been recognised to be in the core of
meaning of the relevant words, then all that is left is to performa
syllogism.11This is so because when the relation "rule-world" been
establishedthen has a relation between rule-rule has to be
established, i.e. a relation between a generalrule (like "it shall
be a misdemeanour, punishable fine of ?5, to sleep in any railway
station")and a particular one by ("the defendantshould pay ?5") has
to be established.Logic does not answer the question of whether a
Cadillac is a vehicle; this is somethingthatfollows from the very
meaningof "vehicle",in such a way thatnot to see this is to show
plain ignoranceof English. But once thatquestionis answered,logic
(in the positivistview) mustbe able to answerthe questionof
whetherthatCadillacis to be allowed in the
park.12significantdifferencein the way in which logical languageis
used and lawyers:for the latter"syllogism","deduction" "logic"are,
and by logicians while for the latterthey arequite
different(however broadlyspeaking,synonyms, related)things. See
Kneale and Kneale (1962). I will follow the lawyers' usage. 12 To
see thatMarmor'sclaim thatHartdoes not "subscribe the view thatthe
to of legal rules is a matterof logical inference"(Marmor1994, p.
128) application is simply false all one needs to do is to read the
passages in which Harttalked abouthardcases, in orderto see what is
theirimplicationfor clear cases: "human inventionand natural
variants processescontinuallythrowup such ["penumbral"] on the
familiar ["core"],and if we are to say that these ranges offacts do
or do not fall under existing rules, then the classifier must make
a decision which is not dictated to him [... ]" (Hart 1958, p. 63);
"If a penumbraof uncertainty must surround legal rules, then their
applicationto specific cases in the penumbral all area cannot be a
matterof logical deduction, and so deductive reasoning [... ]
cannot serve as a model for whatjudges, or indeed anyone, should do
in bringing 11 Thereis a
552
FERNANDOATRIA
A Theoryof Legal Argumentation a Theoryof Legal Reasoning? or
Neil MacCormickbegan his Legal Reasoning and Legal Theory
of witha forceful for argument theimportance syllogistic
reasoning in law,thatis, for the idea thatmodus in alonecan render,
ponens Withthisclaimhe faced somecases,fullyjustified
legaldecisions. thechallenge those(he didnotgive references
thispoint)who of at wouldliketo denythis:If this denial [of the
possibility of legal reasoningbeing deductive] is intended in the
strictestsense, implying that legal reasoningis never, or cannot
ever be, false. It is solely deductivein form,thenthe denialis
manifestlyanddemonstrably sometimespossible to show conclusively
that a given decision is legally justified by means of a
purelydeductiveargument(1994, p. 19).
The importanceof this claim shouldbe by now evident.If it can be
shown that in some cases at least legal reasoningcan be solely
and
to in strictlydeductive form,thenall thatwill remain be doneis
to do in chapter of his book)the 3 tries specify(as MacCormick to
Oncewe know andlimitsof deductive reasoning. presuppositions
wouldbe freeto saythatthose and thesepresuppositions limits,we fail
casesin whichsomeof thosepresuppositions (orthosecasesthat
arebeyondsuchlimits)arehardcases, wherethereis no difficulty of at
all to acceptthatthe question whatthe law is for the case (or
better: be) canbe linkedto thatof whatthelaw oughtto be for will if
it. This is the reasonwhy MacCormick's argument, successful, of
couldbe usedto defenda positivist theory lawlike Hart's. with some
detailit Beforeexamining MacCormick's argument wouldpay,I
believe,to pausefora whileon whatprecisely is that it was is whenhe
saysthattheDanielsdecision MacCormick claiming in manner. justified
a deductive thesis is open to an This is important
becauseMacCormick's that interpretation wouldmakeit
trivial.Indeed,we shall see that his MacCormick himselfsometimes
seemsto understand argument in thisway. of Fora start, consider
Robert Alexy'stheory legalinterpretationas set out in his A
Theoryof Legal Argumentation (1989). In it, heparticularcases under
general rules. In this area men cannot live by deduction
alone"(ibid., at 64; all the italics here are mine).
LEGALREASONINGAND LEGALTHEORYREVISITED
553
from "external" begins by distinguishingwhat he calls "internal"
justification:Legal discoursesareconcernedwith thejustificationof a
specialcase of normative statements,namelythose which expresslegal
judgments.Two aspectsof justification can be
distinguished:internaljustificationand
externaljustification.Internal justificationis concernedwith the
question of whetheran opinion follows logically fromthe
premissesadducedas justifying it; The correctnessof the premisses
is the subject-matter the externaljustification(Alexy 1989, p.
221). of
For Alexy, the problemof internaljustificationis that of
deductive reasoning: "problems associated with internal
justification have been widely discussed underthe heading 'legal
syllogism'" (Alexy 1989, p. 220). Now the importantpoint here is
that no decision is fully justified if it has not been externally
and internallyjustified. For the
externaljustification,non-deductivereasoningis typically needed.
Once the premisses have been (externally)justified (using whatever
criteria is used to justify premisses: consequential reasoning,
purposive interpretation,authority reasons, etc), then it is
possible to say that the decision is fully justified if it follows
in a formally valid mannerfrom those (externally)justified
premisses.13 Notice that for Alexy (unlike MacCormick)the
requirementof thejustificationbeing deductivehas nothingat all to
do with the fact of the case in which it occurs being clear or
hard. The difference will usually lie on the fact that the
(external)justification of the premisses will normally be more
controversialin hard cases than in clear ones; but however
controversialthe externaljustification of the premisses is, once
they have been justified, then the internal justificationtakes over
in the same way for one case or the other. aimed at Thus, in the
context of a theory of legal argumentation, establishing"how fully
to justify a legal judgement"(Alexy 1989, p. 2),
deductivereasoningis to be used in every case.Example: In Hart's
case of the electrically propelled toy car and the "no vehicles in
the park"rule, the premise "a toy car is a vehicle" would have to
be justified accordingto the requirements externaljustification.But
once that of question is settled, all that is left is to deduce
from the statement(thusjustified) "this toy car is a vehicle" and
the rule "no vehicles in the park",the conclusion thatthis toy car
is not allowed into the park. 13
554
FERNANDO ATRIA
We can now see how MacCormick'sthesis can be trivialisedby
takingit to meanonly thata formof deductivereasoningis somehow for
important legal reasoning.In any case syllogistic reasoningcan play
a part. To see this imagine the motherof all hard cases, then
settle (accordingto your moralor legal intuitions)the controversial
aspects of it and on you go! You are now ready to solve the case
with "syllogisticreasoningplaying a role". These arethe reasonswhy
I believe this is not a correctinterpretation of MacCormick'sclaim.
But if this interpretation incorrect, is To thenhow are we to
understand MacCormick's argument? answer this questionwe can turnto
JosephRaz's distinctionbetween what he calls the "narrow" the
"wide"versions of the sources thesis and (Raz 1985, pp. 214-215).
The wide sources thesis "claimsthat the truthor falsity of [pure
and applied] legal statementsdepends on social facts which can be
establishedwithoutresortto moral argument"(Raz 1985, p. 214). In
these cases, all that is needed to solve the case is to find the
applicablerule(s), and establishthe relevant facts, while the
narrowsources thesis is silent concerningapplied claim, as his
analysis legal statements.I believe thatMacCormick's of Daniels
makes clear, is precisely that sometimesthe justification of a
legal decisioncan be purelyandwholly deductivein form,andit can be
presentedas a syllogism which featuresas majorpremisses only legal
rules (and as minor premisses only statementsof fact): "all of the
majorpremisses involved in the argument[in Daniels], not all of
which were expressly stated, are rules of law for which can
contemporary authority be cited"(MacCormick1994, p. 29) or, as he
claimsjust a couple of pages below,It will be observed that in the
above analysis of the argumenteach stage in the argumentis a valid
hypothetical argumentthe premisses of which are either statementsof
propositionsof law which at the materialtime were true for legal
purposes,or findingsof fact which are also for legal
purposestakento be true,or intermediate conclusionsderivedfrom such
premisses(MacCormick1994, p. 32; my italics).
Thus MacCormick'sargumentis not one aboutwhat makes a legal
justificationa good and completeone, as Alexy's was, but aboutthe
existence of some cases that can be solved in a deductivemanner
using as premisses only statementsof propositionsof law and
find-
LEGALREASONINGAND LEGALTHEORYREVISITED
555
ings of fact. About,I believe we could now say, the truthof the
wide, and not only the narrow,version of the sourcesthesis.14
Actually, later in the book MacCormickseems to acknowledge that in
the first sense (judicial syllogism as internaljustification)
"moments"of deductivereasoningexist even in hard cases, which are
characterised the fact that "deduction comes in only afterthe by of
the argument,settling a ruling in law, has been interestingpart
carriedout"(MacCormick1994, p. 197). In Alexy's terms,the
internaljustificationstartsoff only afterthe
externaljustificationhas taken place, since only after the external
justification (what MacCormickat 197 calls "settling a ruling in
law") the major premisses to be used by the internal justification
will be found. MacCormick's claim in chapter 2 of Legal Reasoning
and Legal Theory,then, amountsto saying that in some cases no
externaljustificationis needed beyond that provided by whathe calls
"thefundamental "thoushalt judicial commandment": not
controvertestablishedand binding rules of law" (MacCormick 1994, p.
195). These are the cases that in jurisprudential jargon are called
"clear"cases, the cases that Hartdistinguishedon the basis that in
them, rules can be applied without courts being required to make
what he called "a fresh judgment"(Hart 1994, p. 135): I take
"withoutthe need for freshjudgement"to mean here "without premisses
needing external justification (beyond MacCormick's judicial
commandment)".15 This mightseem an instanceof labouringthe obvious,
andindeed I thinkit is. My only justificationfor it is
thatMacCormickhimself sometimes equivocatesbetween the trivial
(deductionhas a role to play in legal justification) and the
important(some cases can be decided following a strictly
syllogistic line of reasoning) claims. ITo go back to Hart'sexample
(supra,n. 13): if I want to drivewith my FIAT Regattathroughthe
parkin orderto enjoy the view while I am driving(or to save a
couple of minutes from my journey,etc), it would still be true
that, in Alexy's termsthe premise"thisFIATRegattais a vehicle"calls
for (external) justification. But the whole point of Hart's
distinctionbetween core and penumbrawas that any challenge to the
statement:"thisFIATRegatta... etc" would be regardedas lack of
masteryof the word "vehicle",to clarify which only conceptualor
verbal considerations(if any) are useful. 15 For the meaning I am
ascribingto the phrase "freshjudgement"see supra, at 7.14
556
FERNANDOATRIA
will come back to this point lateron in this article(infraat
18f), but for the time being suffice it to comparethe two following
statements by MacCormick:[S]ome people have denied that legal
reasoning is ever strictly deductive. If this denial is intended in
the strictest sense, implying that legal reasoning is never, or
cannot ever be, solely deductive in form, then the denial is
manifestly and demonstrablyfalse. It is sometimes possible to show
conclusively that a given legal decision is legally justified by
means of a purely deductiveargument (MacCormick1994, p. 19, my
italics). [D]eductivereasoningfrom rules cannot be a
self-sufficient,self-supporting, mode of legal justification. It is
always encapsulatedin a web of anteriorand ulteriorprinciplesand
values, even thougha purelypragmaticview would reveal
manysituationsandcases in which no one thinksit worththe troubleto
go beyond the rules for practicalpurposes(MacCormick1994, p. xiii,
my italics).
Daniels v Tarbard We are now ready to examine
MacCormick'sexample of a case in which a purely syllogistic
justificationof the decision is possible. His example was Daniels
and Daniels v R. White & Sons and Tarbard (1938 4 All ER 258).
Though MacCormickhas made this case famous, it seems appropriate
give a brief descriptionof its to facts:Mr Daniels boughta bottle
of lemonade(R White'slemonade) in the defendant's(i.e. Mrs
Tarbard's) pub.He took the bottlehome, where he and Mrs Daniels
drankfrom it. As a consequence, they both became ill, because (as
was proven later) the lemonade was with carbolic acid. Mr and Mrs
Daniels sued heavily contaminated the owner of the pub and the
lemonade's manufacturer. While the latter was absolved from
liability, the former was held liable and orderedto pay damagesto
the (first)plaintiff.MacCormick'sclaim is thatthe court'sdecision
follows in a deductivemannerfrom these facts plus the legal rules
as they were in 1938. could As a matterof fact (of logic, rather),
however,MacCormick not have shown that the court's reasoning in
Daniels was strictly deductive without using the relationshipof
material implication, "D". "D" is used instead of "if in any case
...then ..." (1994, p. 29).16 But legal rules do not rule for all
cases, even if their16 MacCormick probablydoes not mean
materialimplicationin its technical
sense. In symbolic logic, (p D q) "is trueif "not-por q" is
true.But "not-por q" is
LEGALREASONINGAND LEGALTHEORYREVISITED
557
language may induce one to think they do. They do not rule "in
all cases, if... then ... ", but "if in normalcases ... then ... ".
This controversialagainstMacCormick, point should not be
particularly who has always believed that legal rules rule for
"normal" cases, the establishingwhat is to be "presumptively" case
(MacCormick MacCormick 1974: 71; see also MacCormick1995).
Furthermore, explicitly rejects in his book the move made by some
authors,of explaining defeasibility on the basis of moral
disagreementabout the issue of whether or not the law should (moral
"should")be applied. He thinks that in those kinds of cases what is
at issue is not whetherthere are moral reasons to breakthe law, but
what the law actuallyis:[A] positivisticdescriptionof the system as
it operatescannot answera particular kind of questionwhich may be
raisedinternallyto a legal system: the questionas it mightbe
raisedfor a judge in a hardcase: "Whyoughtwe to treatevery decision
in accordancewith a rule valid by our criteria of validity as being
sufficiently justified? and that is a question which can be, and
from time to time is, raised [...]. For my partI should be
reluctantto treatsuch questionsas being non legal as simplybecause
of a definitionalfiat[...]. To treatsuch arguments
ideologicalbut-not-legal(which is what Kelsen and, in effect,
Hartdo) on a priori grounds seems to me unsatisfactory"
(MacCormick1994, p. 63; only the fourthitalics are mine).
as To put it in the wordsused above:if rules areunderstood
referring to normalcases, then they simply cannotbe
appliedwithouthaving previouslyestablishedthat the case is normal.
It is still possible to say (with Kelsen and Hart) that as a matter
of law all cases are normal (or, what amounts to the same thing,
that legal rules are, accordingto the law, to be appliedto all,
insteadof normal,cases), but this implies a definitionalfiat that
begs the question: the fiattruein any one of the following cases:
(1) p is trueand q is true;(2) p is false and q is true;(3) p is
false and q is false [... S]o long as p is false, no matterwhatq
is, "pimplies q" is true;and so long as q is true,no matterwhatp
is, "qis impliedby p" is true (Cohen and Nagel 1934, p. 127). This
is because "materialimplication is the name we give to the fact
that one of a pair of propositionshappensto be false or else the
otherhappensto be true"(ibid. at 128). But MacCormickwants to say,
I believe, that (p D q) means something else, to wit, that because
of p then q. MacCormickmentions this problem, and claims that
"nothingturns on that"(MacCormick1994, p. 28n). I take him to be
offering an stipulationof the meaningof "D", so thatit means "if in
any case p, then (because of p) q" (notice the important in any
case"). "if
558
FERNANDO ATRIA
of saying that according to the law legal rules are to be
applied to all cases (or that according to the law all cases are
normal), however absurd the result might turn out to be. Only after
this fiat will the decision not to apply the law because of these
absurd one. outcomesbecome an "ideological-but-not-legal"
MacCormick is reluctantto endorse this solution, and hence he is
committedto claim that, as a matterof law (and not as a matterof
ideology or morals) legal rules apply to normalcases (indeed, this
is the view thatMacCormick presentlyendorses:cf. MacCormick1974,
1995). But if MacCormickaccepts that laws are to be understoodas
referringto normal, instead of all, cases, then it is difficult to
see how can he claim that that the decision in Daniels was strictly
and liable "withsome regret, solely deductive.Lewis J held Mrs
Tarbard who is a perfectly innobecause it is ratherhardon Mrs.
Tarbard, cent person in the matter"(cit. in MacCormick 1994, p.
21). He for thoughtthe applicationof the law to be inappropriate
the case. It is easy to see why: Lewis J assumedthat in a civil
liability case it is normally the case that if the defendantis "a
perfectly innocent person in the matter" judgement should not be
passed against him or her. In other words, the "innocence"of the
defendant is usually a relevantsubstantiveconsideration.Because in
the court's the understanding rules excluded this
consideration,their application to this particularcase producedsome
inappropriateness: they demandedjudgement to be passed against a
"perfectly innocent was not, in Lewis J's view, person". But this
inappropriateness importantenough for the need for predictabilityto
be waived.17 In other words, he took the rules as being formal
enough to of trump the inappropriateness finding against a
"perfectlyinnocent"party,this considerationnot being strongenough
to make the case "abnormal". This "freshjudgement"was, for Kelsen
and Hart (as MacCormicksays) not requiredby the law: it was
"ideologicalbut-not-legal".But MacCormicksensibly rejects this
position as based upon a definitional fiat that effectively begs
the question.It must be borne in mind that I have legislated above
(at n. 1) the meaning of "predictability", such a way thatit
encompassesall the values that standfor in a
formalisticapplicationof a legal rule. Predictabilityin its
non-stipulated sense is normallythe most important them (hence the
stipulation),but it need not be of the only one.17
LEGALREASONINGAND LEGALTHEORYREVISITED
559
Hence, for MacCormickthis "freshjudgement"is legal, i.e. what
the law is for the case cannot be known before it is made.
Therefore MacCormick'ssyllogism will not be formally valid unless
it is stated as a premise. This can clearly be seen when
attentionis translation the court'sdecision into logical of paid to
MacCormick's notation: (16) If a seller has broken a condition of a
contract which he was requiredto fulfil, the buyer is entitled to
recover damages from him equivalent to the loss directly and
naturallyresultingto him from the seller's breach of the condition;
In the instantcase, the seller has brokena conditionof the
contractwhich she was requiredto fulfil; .'. In the instant case,
the buyer is entitled to recover damages from her equivalent to the
loss directly and naturallyresultingto him from the seller's breach
of the condition(MacCormick1994, pp. 31-32).18
(15) (17)
This is translatedas (the left column is MacCormick's,while the
of rightone containsmy translation MacCormick'slogical notation
back to English, according to his stipulationson pp. 23 and 28f,
which I will use thereafter): (16) (15) (17) y Dz y .'. z. (16) In
any case, ifythenz; (15) In the instantcase, y; (17) Thereforein
the instantcase z
MacCormickis clearly correct in claiming that (17) follows from
(16) and (15). But the point is that (16) is not a correct
description of the law as it was at the time, and we have already
seen18 MacCormick's complete syllogism is considerablylonger (cf.
1994, p. 30ff). The objection I am presentingnow could, however, be
directed against any of its parts, therefore it is enough for me to
quote a section of the reasoning. It is also worth noticing that
though MacCormicknow believes that a judicial syllogism like
Daniels's should be representedusing predicateratherthan
propositionallogic, I have retained MacCormick'soriginal
representationof it (see MacCormick1994, p. xv; MacCormick'schange
of mind was promptedby White 1979).
560
FERNANDOATRIA
that MacCormickelsewhere in the book (and in other writings,
most notably, 1974, 1995) agrees with this. If we correct (16) by
the introducing idea of "normalcases", we would get (16') (15) (17)
In normalcases, if y, then z; In the instantcase, y; Thereforein
the instantcase, z.
And this is not a valid deductive argument:to be one it needs a
furtherpremise: (18) The instantcase is a normalcase
MacCormick'spreferredoption (that legal rules establish what is
to "presumptively" be the case) makes this problem even more
noticeable.For consider: z; (16") If y, then presumptively (15) In
the instantcase, y; (17) Thereforein the instantcase, z. Again,
(17) does not follow. Whatdoes follow is z. (17") Thereforein the
instantcase, presumptively But (17") does not, of course,justify a
legal decision. It does not tell anybodywhat the law is for the
instantcase: it only states what the law "presumptively" is. What
MacCormickcalls "the pragmaticsof law" (1994, p. xiii; 1995) would
not be of much use here. "Arule thatends with 'unless ...' is still
a rule", of course, but it cannot be applied unless the is
exceptionalcircumstance not present.The rule mightbe such that the
"default" position is that the exception does not exist, but even
in this case the justification would, from a logical point of view,
be incomplete (i.e. invalid)if this circumstanceis not asserted.For
consider,
LEGALREASONINGAND LEGALTHEORYREVISITED
561
(16'") (15'") (17'")
In any case, if y, then z, unless the court is satisfiedofw;
In the instantcase, y; Thereforein the instantcase, z.
Again, (17"') fails to follow. For the argumentto be
formallyvalid, a premiselike the following is needed: (19) w has
not been proven (or: "the court has not been satisfiedof w").
Following Hart, we have already seen that "w" here stands for of
a fresh judgement to the effect that the inappropriateness the
applicationof the rule to the particularcase is importantenough to
for the demandfor predictability be waived. As a premise,therefore,
(19) is neither a rule of law nor a statementof fact, but an
evaluativejudgement:"in this case the result offered by the rule is
or not inappropriate, at least not to a significantextent". In
other even in as clear a case as Daniels and even assuming that
words, the court has the obligation to apply the law, no decision
can be reached in a syllogistic mannerusing only rules of law and
statements of fact as premisses. The fact thatthe absence of w need
not be argued, importantas it is from a pragmaticpoint of view (no
externaljustificationis needed to regardit as absent)is immaterial
from a logical point of view.19 In short, the only way in which
MacCormick's claim could succeed is assuming the definitionalfiat
he (rightly, in my view) rejectsin chapter3.19 Cf. MacCormick 1994,
29, where MacCormick rightly points out that p. to the premisses
stated by Lewis J a furtherone should be added, one "which is so
trivially obvious that its omission from the express statementsof
Lewis J is scarcely surprising- namely that the
transactiondescribedin (i) above was intendedby each of the
partiesto be a purchaseby Mr.D. FromMrs. T. and a sale by her to
him".Maybe the premisethatstatesthe normalityof the instantcase (or
thatthe presumptionin favourof the solution offered by the rule
accordingto its meaning is not defeated in the instantcase) is
equally trivially obvious in many cases, but as MacCormick
recognises the fact thata premiseis "triviallyobvious" for does not
meanthatit is not required the formalvalidityof the
inference,though it might very well mean thatthe courtis
justifiedin not statingit.
562
FERNANDO ATRIA
Now, it could be arguedthatI have missed the point, thatthe fact
case at hand is one that the rule should be appliedto the
particular of the presuppositions(and it thus constitutesa limit)
of deductive justification.In Legal Reasoning and Legal
Theory,MacCormick of says thatone of the presuppositions legal
reasoningis thatevery judge has in virtue of his office a duty to
apply each and every one of those rules which are "rulesof law"
wheneverit is relevantand applicableto any case broughtbefore him.
And that formulationreveals a second presupposition, withoutwhich
the term"duty" would lack identifiable reference:thatit is possible
for the judge to identify all those rules which are "rules of law"
(MacCormick 1994, p. 54).
Hence, the counter-objectionwould continue, if it is doubted
whetherthe rule should be applied to this particular case, then we
are going beyond one of the limits of deductive reasoning, while
MacCormick'sthesis was meant for those cases in which those
presuppositionsare satisfied. But this cannot be an answer to my
claim, since I am assumingthatthe court objectionto MacCormick's
has to applythe law; what I am contestingis thatin findingwhat the
law is for the case, the court will necessarily have to assume that
if the case is "normal" rules like those in Daniels are to be
applied as they were in that case. This, again, could be used to
defend MacCormick'sposition only if one were to adoptthe solution
that MacCormickis reluctantto adopt,i.e. if one were to claim that
the rule applies to all cases as a matterof law, howeverjustified
(from an "ideological-but-not-legal" point of view) the court might
be in not applying it to the particularcase. Only given that
assumption MacCormickcould say that the process of finding a
solution is (or can in some cases be) deductive:given the
relevantrules as they were in 1938, and the facts of Daniels as
they were provenin court, the conclusioncould be reachedin a
deductivemanner. the same By token, however,he would have to say
that given Pufendorf'sreport of the Bolognese law (and the facts as
he told them), we could reach the conclusionthatthe barberhad to be
punishedin the same deductivemanner. Whatwe would addin the
lattercase wouldbe an to argument the effect thatpunishingthe
"ideological-but-not-legal" barberis too absurdfor the courtto do
it. MacCormick'sargument cannot succeed without this a priori
distinctionbetween the legal andthe ideological, a
distinctionthathe himself thinksis unjustified.
LEGALREASONINGAND LEGALTHEORYREVISITED
563
Since MacCormickhimself rejects this distinctionwe need not
discuss it here.20What interestsme here is to point out the
incompatibility of MacCormick'slegal theory with his account of
legal We reasoning.21 know thatLegal Reasoning and Legal Theorywas
meant to be a Hartianexplanation of legal reasoning. Hence, it had
to claim that some cases were in a Hartiansense clear, that is,
their outcome could be determinedaccordingto the rules alone (that
is the gist of Hart's criticism of rule-scepticism). If those cases
are completely determinedby the rules, it must be possible to
reconstruct justificationof a solutionto them accordingto the the
deductivemodel. Thatis to say: if it is the case thatthe life of
the law consists to a very large extent in the guidanceboth of
officials and private individuals by determinaterules which, unlike
the applicationsof variablestandards, not requirefrom them a
freshjudgementfrom case to case do (Hart1994, p. 135)
then in those cases the court's decision can be representedin a
syllogistic way, in which the only presuppositionneeded (along with
statementsof fact and of legal rules) is that the law ought to be
applied, in which no premise containinga "freshjudgement"is needed
for the formal validity of the inference. This is the significance
of MacCormick'sargumentin Chapter2 of Legal Reasoning and Legal
Theoryas an analysis of clear cases accordingto Hart. But in the
following chapters,in which he undertookto build up a theory of
legal reasoning, he was driven to positions which are
incompatiblewith the claims of the (legal) theory. Thus, when
discussingthe issue of clear and hardcases, he starts by noticing
that "in truth there is no clear dividing line between clear cases
and hard cases" (MacCormick1994, p. 197). There is a spectrumof
cases, ranging from the hardestto the clearest, and20 See my
"Gamesand the Law"cit. supraat n. 4. 21 I am legal theoryas it can
be found in Legal referringhere to MacCormick's Reasoning and Legal
Theory.His position is nowadaysdifferent:"[I] no longer accept
nearly as much of his [i.e. Hart's] theses about law as I did in
1978" (1994, p. xv). My own comments aboutLegal Reasoning and Legal
Theoryare not to be seen as a criticismof MacCormick'slegal
theory,since (I would claim) his later work can accommodatemost of
the claims made here, but about the tension between the
perspectivesof legal theory and legal reasoning, a tension
thatpermeateshis argumentas originallypresentedin 1978.
564
FERNANDOATRIA
acrossthatspectrum"it could neverbe judged more thanvaguely at
doubtscould become significantenough what point"interpretative
court to have discretion.Now instead of offering (like Hart for the
a with his open texture thesis in its first interpretation)
value-free test to distinguisha clear from a hardcase, he finds the
explanation cases in "differat of this uncertainty the divide
between clear/hard in the dominantstyle of differentperiodsin the
historyof legal ences systems"(1994, p. 198). Lateron we are told
that"whenwe talk of differencesbetween judicial styles [... ] what
we are talking about is or includes the degree of readiness which a
judge manifests to that [i.e. permitthatpresumption the presumption
"obviousmeaning to shouldbe preferred"] be overridden" (1994, p.
207). In this view, how pressing the absurdityof the result
produced case shouldbe for the by the applicationof the rule to the
particular in judge to permitthe presumption favourof the obvious
meaningof the wordsto be overridden not somethingthe rule can
settle;it is a is and problemgeneratedby the conflictingdemandsof
predictability a case cannotbe decidedbeforedecidingwhetherit
appropriateness; will be treatedas a "normal" case (and given - and
excluding- this decision a deductivejustificationcould be
reconstructed) as one or in which substantive considerationsshow
thatthe case is abnormal, thatis, is one in which the presumption
must be overridden. To emphasise:if what makes a case clear
ratherthan hard (and vice-versa)is ajudgmentaboutthe
rightbalancebetweentwo values (i.e. a freshjudgment),then at least
some (I would say: all, but all I to need for the argument standis
to say "some")hardcases are hard because they ought to be so. The
only reason,I submit,why MacCormick thinkshe can claim both that
the decision in clear cases can be justified in a syllogistic
manner (using as premisses only statementsof fact and of legal
rules) and thatrules apply only to normalcases (or thatthey
establish only what is to be "presumptively" case) is that he (as
we the betweenthe two differentclaims identified alreadysaw)
equivocates above concerning what we could call the
"deductiveelement" in legal reasoning. MacCormick'sargumentwas
originally presentedagainst those who held the thesis that "legal
reasoning is [n]ever strictly deductive"(1994, p. 19). We are told
thatif this denial "is intended
LEGALREASONINGAND LEGALTHEORYREVISITED
565
in the strictest sense, implying that legal reasoning is never,
or cannot ever be, solely deductive in form, then the denial is
manifestly and demonstrablyfalse. It is sometimes possible to show
conclusively that a given decision is legally justified by means of
a purely deductive argument"(1994, p. 19). Later in the book,
however, chapter 2 was supposed to have been directed against
"thosewho deny that deductivelogic is relevantto the justification
of legal decisions" (1994, p. 45), and in the new forewordto the
1994 paperbackedition the argumenthas definitely changed:now
denials by learned persons that it is presented against "recurrent
the law allows scope for deductivereasoning,or even logic at all"
(1994, p. ix). In the same piece MacCormickseems to reject his own
claim that "it is sometimes possible to show conclusively that a
given decision is legally justified by means of a purely deductive
when he now claims that"deductive reasoningfromrules argument"
cannot be a self-sufficient,self-supporting,mode of legal
justification. It is always encapsulatedin a web of anteriorand
ulterior reasoning from principles and values [...]" (1994, p.
xiii; all the are italics in this paragraph mine). In my opinion,
the quotations from the new foreword reflect MacCormick's present
view of the "centrality of deductive in reasoningfor legal
reasoning"andthey have to be understood the light of Alexy's
distinctionbetween externaland internaljustification. So
understood,the claim refers to the possibility of translating a
given decision in syllogistic terms as being usually the clearest
and safest way to check whetheror not the decision was fully
justified, whetheror not issues requiringexternaljustificationhad
arisen (and if they had, whetheror not they were settled
accordingto the of requirements the externaljustification). in this
sense chapter 2 does not answer the challenge to But legal
positivism it was designed to answer. If it is to provide an
answer,it has to be taken as meaning that sometimes it is possible
for legal decisions to be fully justified througha syllogistic
chain of reasoningthat uses only statementsof fact and of legal
rules as premisses. Only in this sense the thesis would imply, if
correct,the rejection of the argumentpresentedup to now. Only in
this sense could it help Hart to show that in some cases no fresh
judgement is needed for courts and officials to apply the rules.
But for this
566
FERNANDO ATRIA
to argument work,an a priori distinctionhas to be madebetweenthe
and the ideological. Since MacCormickis unwilling to make legal
this ad hoc distinction,the argumentfails to prove that sometimes
and legal decisions can be "purely" "solely"deductivein form.
The reasonsfor consideringin some detailMacCormick's argument
were, as statedabove, not only concernedwith the intrinsicvalue of
it; it also helps us illustratethe contemporary predicamentof legal
theory: dependingupon the perspectiveadopted at the beginning, one
can reach, following naturaland plausible steps, incompatible
conclusions. When MacCormickadopted the perspective of legal
theory, that is, the perspective of an enterprisedirected to
understandingwhat law is, when a legal system exists and the like,
he was drivento the Hartianview that sometimesrules are there, so
to speak, and can sometimesbe "straightforwardly" applied. When he
adopted the perspective of legal reasoning, that is to say, one
that tries to understandhow the law is applied (to my knowledge,
his book is still one of the few, not to say the only one,
self-avowedly positivist work in which the discussion of decisions
given in actual cases plays a crucialmethodologicalrole) he could
not live up to that: the conclusions for legal reasoning that would
follow from the "legaltheory"thesis arejust too implausible,too ad
hoc. I wantto claim thatthis is not a problemof MacCormick'salone.
This problemappearsin one way or anotherin the work of many of the
most sophisticatedauthorswritingtoday on legal theory.I want to end
this article by showing that this is also the case concerning the
place of legal reasoningin JosephRaz's recentwork.LEGAL REASONING,
RULES AND SOURCES
Contraryto what one could guess, Raz believes that "commitment
to the sources thesis does not commit one to formalism or to the
autonomyof legal reasoning"(Raz 1993, p. 317).22Raz here means by
'formalism" the thesis that "the art of legislation, and more
generally law-making, is that of moral reasoning. But legal
reasoning is22
LEGALREASONINGAND LEGALTHEORYREVISITED
567
In 1985, however, Raz presentedan argumentthat could easily lead
one to believe thatthe sourcesthesis impliedindeed some form of
autonomyfor legal reasoning.There Raz distinguishedbetween of two
interpretations the sourcesthesis:Let us distinguish between what
source-basedlaw states explicitly and what it establishes by
implication. If a statute in country A says that income earned
abroadby a citizen is liable to income tax in A, then it only
implicitly establishes that I am liable to such tax. For my
liability is not stated by the statutebut is inferredfromit (and
some otherpremises).Similarly,if earningsabroadare taxed at a
differentratefrom earningsat home, the fact thatthe proceedsof
exportsales are subjectto the home rate is implied ratherthan
stated.It is inferredfrom this statuteand otherlegal rules on the
location of varioustransactions. The two examplesdifferin thatthe
statementthatI am liable to tax at a certain rateis an appliedlegal
statementdependingfor its truthon both law andfact. The
statementthatexportearningsare taxed at a certainrate is a
purelegal statement, depending for its truth on law only (i.e. on
acts of legislation and other lawmaking facts). The sources thesis
as stated at the beginning can bear a narrow or a wide
interpretation. narrowthesis concernsthe truthconditionsof pure The
statementsonly. Pure legal statementsare those which state the
content of legal the law, i.e. of legal rules,
principles,doctrines,etc. The wide thesis concernsthe
truthconditionsof all legal statements,including applied ones. It
claims that the truthor falsity of legal statementsdependson social
facts which can be established withoutresortto moral argument(1985,
pp. 214-215).
Using Raz's language, legal reasoning is reasoning about which
appliedlegal statementsare true (or valid). Underthe
narrowinterpretation, therefore, the sources thesis is silent
concerning legal reasoning, since it is silent concerning applied
legal statements. Since in 1993 we find him categorically saying
that the sources thesis does not commit one to the autonomyof legal
reasoning,we would be forced to understand as an endorsement the
narrow, that of of as opposed to the wide, interpretation the
sourcesthesis. But actually, for Raz the wide and the narrow
versions of the sources thesis stand together, at least if moral
facts are not contingent:All the argumentsso far concern the
narrowsources thesis only. Nothing was said aboutits applicationto
appliedlegal statements.I tend to feel that it applies to them as
well, since they are legal statementswhose truthvalue depends on
reasoning about the law as it is. As such it is free from any
infection by moral reasoning.One can reason morallyaboutlegal
reasoningbut not in it, not as part of it" (Raz 1993, p. 314).
568
FERNANDOATRIA
contingentfacts as well as on law. If one assumes
thatcontingentfacts cannotbe moralfacts, then the sourcesthesis
applies here as well. Thatis, what is required is the
assumptionthat what makes it contingentlytrue that a person acted
fairly on a particular occasion is not the standardof fairness,
which is not contingent, but the "brute fact"thathe performeda
certainaction describablein value-neutral ways. If such an
assumptionis sustainablein all cases, then the sources thesis holds
regardingappliedlegal statementsas well (1985, p. 218).
In brief:the narrowversion,togetherwith the claim thatmoralfacts
are not contingent, imply the wide version and the wide version
implies the autonomy of legal reasoning. Since Raz wants to say
today that the sources thesis does not imply the autonomyof legal
reasoning it would seem as if he owes us an explanationof how
moralfacts are contingent. I believe, in fact, that the
narrowversion cannot fulfil the role Raz expected the sources
thesis to fulfil. Indeed, what are we to make of Raz's
statement(1985, p. 218) that "all the argumentsso far concern the
narrow sources thesis only"? The "argumentsso far"were advancedto
claim that only if the law complies with the sources thesis can it
have authority.The reason for this was that authority-capacity
requiredthe two non-moralfeaturesof authoritative directives,and
thatthey in turnrequiredthe sources thesis. Let directivescan us
focus uponthe second feature,i.e. thatauthoritative be
identifiedand their content ascertainedwithoutusing evaluative
directivewas considerations(the first one was that an authoritative
to reflect someone's view on the balance of applicable supposed The
answeris: because reasons).Why was this conditionrequired? if it
were not met the would-be directive would fail to be able to fulfil
its function,and subjectswould fail to be able to be benefited by
the existence of the authority:the subjects "can benefit by [the
authority's]decisions only if they can establishtheir existence and
contentin ways which do not dependon raisingthe very same issues is
which the authority thereto settle"(Raz 1985, p. 203). Only if this
directivebe able condition is met would an allegedly authoritative
to comply with the normaljustificationthesis. But at the end of the
so article we are told that the "arguments far"concern the narrow
version only, with the obvious implication that
authority-capacityrequires only the narrow sources thesis. This
means that the argu-
even ment turnsout to be that the law can have
authority-capacity if the wide sources thesis is untenable,that is,
even if subjectscan
LEGALREASONINGAND LEGALTHEORYREVISITED
569
neverget any "appliedlegal statement" withoutraisingall the
moral considerationsthat were pre-emptedby the authoritative
directive, even if the authorityis fully legitimate. In other
words: if we accept Raz's claim that "all the arguments presented
[in the first four sections of 'Authority, Law and concern the
narrowsources thesis only" the authoritaMorality'] tive natureof
law ceases to be an argumentfor the sources thesis. For
consider:Raz claims that "a decision is serviceable only if it can
be identifiedby means otherthanthe considerationsthe weight and
outcome of which it was meant to settle" (Raz 1985, p. 203).
Serviceable for what? For the parties to be able to act upon the
decision rather than their own judgement. But to be serviceable in
these terms what is requiredis the wide sources thesis, i.e. that
is (providedthatthe authority legitimate),subjectscan stop thinking
directiveand aboutthe substantive problembehindthe authoritative If
simply do as it commands.23 this cannotbe done no authoritative
directivecan ever comply with the normaljustificationthesis. But
the distinction between the narrow and the wide interpretations of
the source thesis was not mentionedby Raz in 1993. Is Raz's
"TheAutonomyof Legal Reasoning"a rejectionof it? We are not given
an answerto this question. InsteadRaz offers two differentreasons
why legal reasoning is not autonomousfrom moral reasoning:the first
has to do with the fact that, "if our sole concern is to work out
what ought to be done in orderto obey the intentions,purposesor
goals of the law-makers, we will often find ourselves faced with
conflictingdirectives"(Raz 1993, p. 315). In this case, a choice is
necessary, and the choice cannot be guided by
source-basedconsiderations.It follows that they have to be moral
considerations.24But this in turn is not argumentfor the
compatiblewith the thrustof the authority-based23 NB: the
argumentis silent regardingwhat shouldthe subjectsdo. It does
not
claim thatthey shouldfollow the authoritative directive,only
thatif thatwere the case it would be possible for them to do it. If
it is possible for subjectsto comply with the directives,then this
implies (at least following Raz's theoryof authority) that it is
possible for them to take them as protectedreasons, i.e. as reasons
for action that also exclude otherconflictingreasons. 24 Raz claims
thatnon source-based cannotbutbe moralconsidconsiderations
erations,"forthere is no otherjustificationfor the use of an
autonomousbody of considerationsby the courts"(Raz 1993, p. 318).
Thereforethe question of the
570
FERNANDO ATRIA
was thatif the authoritative sourcesthesis. Recall thatthe
argument directives claim legitimate authority,it follows that they
can have it If authority. they can have authority, follows thatthey
mustposses one the non-moralconditionsfor havingauthority, of which
was that the subjects must be capable of establishingthe
directives' "existence and content in ways which do not depend on
raising the vary is sameissues which the authority thereto
settle"(Raz 1985, p. 203). But now Raz seems to be claimingthatwhen
applyingsource-based materialour "sole concern"is not to applythe
directivesthus recognised, but to decide "what ought to be done in
order to obey the Raz seems to be intentions,purposes,or goals of
the law-makers". claiming that our sole concern should be that of
second-guessing the authority,going beyond the meaning of the
directive to check whetheror not thatmeaningis a
correctreflectionof the authority's "intentions,goals and
purposes".But we should not second-guess if the authority, the
sourcesthesis is true.25 Let me pause for a while on the meaning of
the "should"that appearedin the last sentence. Since we are
consideringwhetheror not legal reasoning is autonomousfrom moral
reasoning,it seems here to understand "should" its moralsense. Last
this in appropriate last paragraph's sentence, so
understood,assumesthatthe authority is legitimate. Needless to say,
in many situations this will not be the case. Sometimes the
authoritywill be a de facto authorityand courts will have no reason
at all to follow its directives.But this is immaterial to the
discussion of the autonomyof legal reasoning, since,autonomyof
legal reasoningcan only be the questionof its autonomyfrom moral
reasoning. 25 Raz could claim here that I missed the point, which
is the fact that the law of conflictingvalues [... ] due to the
fact that [it] is a product displays "plurality of human
activity"(Raz 1993, p. 315n). But consider a case like Fuller's men
sleeping in the station (see Fuller 1958). Here it might well be
the case that thereis a "plurality conflictingvalues"(select the
pair of your choice: predictof ability againstappropriateness,
keepingrailwaystationsclean againstfairness,or but the fact is, the
source-basedmaterialdoes offer a solution: fine whatever), the
first man and acquit the second. The problem created by the
"pluralityof conflictingvalues"will only be seen by the courtif the
courtdoes precisely what it is not supposedto do, i.e. if it
"raisesthe very same issues which the authority is thereto
settle"(cf. Raz 1985, p. 203). No conflictis evidentif the
courtfollows the law as identifiedaccordingto the
sourcesthesis.
LEGALREASONINGAND LEGALTHEORYREVISITED
571
from Raz's point of view, this is a moral question ("how,all
things considered,should the courts decide the case?"), which is
different from the legal question ("how, according to law, should
cases be decided?").The fact that courts sometimes have the moral
duty to authoritative directivesdoes not show anythingaboutlegal
disregard since thatis a questionaboutwhetheror not the law ought
reasoning, to be applied,while legal reasoningdeals with the
questionof what is the law for the case (Raz 1993, p. 312). If this
is the only way in which moralreasoningand legal reasoningare
connected,the latter could still be autonomousfrom the former. But
Raz wants to deny this, since he wants to claim that "legal
reasoning is an instance of moral reasoning".Thereforehe has to
show why legal reasoningis moralreasoning,even when the question of
whether or not the law should (morally) be applied is not taken
into account.To do this he distinguishesbetween "reasoning about
the law" and "reasoningaccording to law". The first (i.e. aboutthe
law") "is governedby the sourcesthesis"(Raz "reasoning 1993, p.
316), hence if we restrictour view to it an autonomousform of legal
reasoningwill appear.But we should not leave the second aspect of
legal reasoning, i.e. "reasoningaccordingto law" aside, and once we
pay attentionto it, Raz tells us, we shall realise that
moralreasoning"(Raz 1993, it is "quitecommonly straightforward p.
317). Thatreasoningaccordingto law is differentfrom reasoning
aboutthe law is shown by thatfact thatThe law itself quite commonly
directs the courts to apply extralegal considerations. Italian law
may direct the courts to apply Europeancommunity law, or
international law, or Chineselaw to a case [... ]. In all these
cases legal reasoning, understoodto mean reasoningaccordingto law,
involves much more thanmerely establishingthe law (Raz 1993, p.
317).
This might be so, but that does not show that legal reasoning is
a form of moralreasoning.The most it could show is thatItalianlegal
reasoningis a form of Europeanlegal reasoning(not thatthis makes
any sense). So let us consider whetherlegal referencesto morality
ratherthanto Chinese law would fare betterfor Raz. Wouldthe fact
thathere andtherea legal systemmay containreferencesto morality
show thatlegal reasoningis a form of moralreasoning? I hope the
answer to this question is evident: insofar as particularrules
makereferencesto morality,then "reasoning accordingto
572
FERNANDOATRIA
law" is more than "reasoningabout the law". But this argumentis
not enough to prove that "legal reasoning is an instance of moral
reasoning"any more that the fact that sometimes engineers should
consider aesthetic considerationsmakes engineering-reasoning an
instanceof aestheticreasoning. In brief, Raz does not want to draw
the implications of his legal theory for legal reasoning.He tries
to show that the sources thesis does not commit one to the thesis
of the autonomyof legal reasoning,and to say so he has to make
space for somethingto be left afterthe existence and contentof the
source-basedmaterialhas been established.In the end, he can only
come up with the small space providedby the fact thatsometimesthe
law instructscourtsto and applyextralegalconsiderations he offers
this as a groundfor the grandthesis that legal reasoningis moral
reasoning.My argument all along has been that this latterclaim is
indeed true, and because this is the case anythinglike the
sourcesthesis cannotbut be false. To make this point clearer,let me
consider a more recent effort by Raz to show thatthe sourcesthesis
does not commit one to what In (in 1993) he called "formalism". "On
the Nature of Law" (Raz 1996) he triedto defend his legal
positivism againstthe chargethat it misrepresents legal
reasoning.The "standard objection",he says, to it when its
implicationsfor legal reasoningare drawn,is thatwould we not expect
two clearly separate stages in legal reasoning: an stage anda
(purely)moralone? Firstone would establishwhat interpretive-factual
laid down law says on the issue at hand, and then either it does
authoritatively not provide a determinatedisposition of the issue,
or if one wants to determine whetherthe way it disposes of the
issue is morally acceptable,one would move to the second
purelymoralstage in the argument. fact we do not find thatlegal In
divides in that way. Legal reasoningdisplays a continuitythroughall
reasoning its stages (Raz 1996, p. 19).26
To show how this objection actually reinforces ratherthan
refutes his
views on the natureof law, Raz invites us to
considerinterpretationincidentallyairs some doubtsas to what he
sees as the tacit assumption of this objection:"I believe that this
point is overstated,