Principles for choice of sentence: Theproposals of Fängelsestraffkommittén




The Committee on Imprisonment (Fängelsestraffkommittén) has proposed a comprehensive reform of Sweden's laws concerning criminal sanctions. To a foreign observer, the most noteworthy of its suggestions are the proposed statutory principles for choice of sentence, to be embodied in Chapters 33 and 34 of the Penal Code.1 It is these principles I shall address in this essay.
    As someone who received his legal training elsewhere, I feel somewhat diffident addressing a Swedish audience about changes in Swedish law. I have, however, participated in the debate here and in the U.S. over the rationale for criminal sentencing, and have tried tofollow recent efforts in various jurisdictions to regulate sentencing choices through law or guidelines.2 I shall try to relate the Imprisonment Committee's proposals to this wider discussion. The Committee's report is not an academic document but a practical reform effortby a group of men and women having varying experience and outlooks. Inevitably, it has its areas of compromise and tension. The proposals offer, nevertheless, a principled way of deciding sanctions and deserve serious consideration not only by Swedish but by foreign jurists.


I. Why have guidance?
Why should there be legal guidance for deciding sentences? The customary answer to this question has been that guidance is needed toalleviate "disparity" in sentences. Without a coherent policy for sen-


* This article was originally presented at a symposium at the Law Faculty of Uppsala University, held on October 18, 1986. The author wishes to thank Nils Jareborg and Lena Holmqvist, as well as the other participants in the symposium, for their comments.

1 SOU 1986:13, s. 76-83.

2 See von Hirsch, Past or Future Crimes; and von Hirsch, Knapp, and Tonry, The Sentencing Commission and its Guidelines. 1-37-161 Svensk Juristtidning


2 Andrew von Hirschtencing, however, disparity cannot even be identified. Disparity, by its usual definition, consists of inequalities of sentence not justifiable by the sentencing purpose being sought. Traditionally, sentencing law has lacked identifiable purpose. It is variously said that the sentence should rehabilitate, deter, or incapacitate offenders, or punish them as they deserve. These aims conflict, however, -and each implies a distinct set of criteria for determining sentence. Without specification ofthe predominant purpose, the criteria remain unknown.3
    Is it disparity, for example, when two people commit the same crime, and the one having the more favored social status gets the morelenient sentence? That surely seems like disparity, but on the above definition it may or not be so. Were one's aim to incapacitate persons likely to return to crime, such divergent sentences might seem justifiable-because employment, income and educational attainment are somewhat correlated with probability of recidivism.4 Were the aim to punish proportionately and deservedly, the difference would constitute disparity, however-as social class does not bear on the gravity of the criminal conduct. What is missing in Swedish law, as in the law of so many other jurisdictions, is a specification of the comparative weight to be given to prediction, desert, and other purposes in sentencing.
    The absence of adequate guidance is evident on even a cursory examination of current Swedish law. The Penal Code (1:7 BrB) provides that the sentences should (1) maintain general obedience tolaw and (2) foster the defendant's rehabilitation. It is far from clear, however, how the courts are to accomplish these potentially conflicting purposes-especially given the paucity of effective treatments and the tenuous connection between general law-abidingness and the sentence in any particular cases. The Code fails to suggest what features of the offender or his offense should ordinarily be given emphasis. And it leaves critically important issues unaddressed, such as the ones just mentioned: How important are concerns about proportionality between the gravity of the crime and the severity of the punishment? To what extent, if at all, should incapacitation influence the sentence?
    Guidance for sentencing is needed, then, to provide a policy: to help choose which sentencing aim should predominate, and thereby to help decide what features of the offender or his offense should be given


3 This issue-the conceptual link between sentencing purpose and the definition of disparity-will be addressed in the forthcoming Eighth Colloquium on Criminology of the Council of Europe, to be held in November 1987.

4 See Past or Future Crimes, chap. 9.


Principles for choice of sentence 3most weight. The Imprisonment Committee's proposals attempt to supply such a policy-one based on ideas of proportionality.


II. How much guidance?
Within the maxima and minima provided by the statutory penalty scales, how much guidance should be provided? The answers to this question have varied. At one extreme, is the English approach of leaving the task of guidance wholly to the appellate courts, without substantial assistance provided by statute or regulation.5 At the other, is the American approach (tried in Minnesota and several other states) of numerical sentencing grids that furnish specific terms or narrow ranges as the normally-recommended sentences.6 The Swedish Imprisonment Committee has opted for a middle way. The legislation is to provide general principles but no numbers. The courts are then to apply those principles in deciding the quantum of sentencein individual cases.
    Leaving guidance to the appellate courts —the English solution—has been defended by the Cambridge jurist, D.A. Thomas.7 Britain's Court of Appeal, Criminal Division, issues opinions when it reviews sentences from lower courts; these opinions, Thomas contends, have coalesced into a case-law jurisprudence for sentencing. No general legislation on choice of sentence is needed, he argues, because the Court of Appeal has matters well in hand. Other British observers, however, do not share Thomas' sanguine view. They question the extent to which an intelligible pattern emerges from the Court ofAppeal's decisions; and also question whether the lower courts, in the majority of cases which are not appealed, give much weight to the Court of Appeal's sentencing opinions.8 The sentencing the oriesenunciated in the Court's decisions lack sophistication, moreover. Claims about deterrence or treatment are made with little regard forthe availability of supporting evidence; concerns about proportionality are not well articulated.9 Given the press of other court business, and judges' limited time and opportunity to peruse sentencing research, such deficiencies are not surprising, but the English sentencing "jurisprudence" does not suggest great promise.
    Thomas' solution of unaided appellate-court guidance does not, inany event, seem suited for Sweden. The Swedish appellate courts—the


5 See Thomas.

6 For a description of such numerical guidelines, von Hirsch, Knapp, and Tonry, chaps. 1 and 2.

7 Thomas.

8 For a fuller critique, Ashworth.

9 See von Hirsch, Guidance by Numbers or Words?


4 Andrew von HirschSupreme Court and the six Courts of Appeal-have for years had authority to review sentences. No comprehensive case-law jurisprudence has emerged, and Swedish appellate courts generally have not emulated the English Court of Appeal's practice of issuing "guideline" judgments for selected types of cases.10 It is not that those courts have been inactive. According to an unpublished recent survey by Alvar Nelson, a surprisingly large percentage of trial-court sentences are modified by the Courts of Appeal each year.11 Those decisions to modify, however, are focused on the particular facts of the case—and seldom offer broader guidance. Since the Penal Law offers the appellate courts no real guiding principles for choice of sentence, the tendency to restrict appellate decisions to the particular fact situation constitutes understandable self-restraint.
    At the opposite extreme, lies Minnesota's solution: the numerical sentencing grid. The grid is a two-dimensional table of prescribed sanctions. Its vertical axis is an offense score, which grades the seriousness of the current offense of conviction. Its horizontal axis, or offender score, grades the extent of the offender's prior criminal record. Across the grid is drawn a so-called dispositional (or "in-out") line. Above the line are prescribed prison terms of varying duration, and below it are lesser sanctions. In each cell in the grid above the line, a narrow range of imprisonment is prescribed: the grid cell applicable to convicted armed robbers having two prior felony convictions might contain a range of, say, 38 to 44 months. Such ranges, however, constitute only the normally applicable sentence. The courtis authorized to deviate from the prescribed ranges on account of aggravating and mitigating circumstances—and the guidelines contain a suggested list of factors that qualify as extenuating or aggravating.12
    The commonly-expressed objection to Minnesota's system of numerical guidelines is that it is merely mechanical—"sentencing by computer". That is not quite accurate. Courts may, as I noted, deviate from the grid ranges for aggravating and mitigating circum


10 There are some exceptions. The Swedish appellate courts have, for example, addressed the sentences generally appropriate for drunken driving. Detailed guidelines for prosecution-imposed penal orders are also issued on a regional basis. For most crimes, especially those of a more serious nature, however, there is no systematic guidance provided. An example of an English "guideline" sentence is R. v. Billiam, [1986] 1 All E. R 985, prescribing five years imprisonment as the "starting point" for forcible rape in contested cases.

11 Professor Nelson reports that in 1985, more than 40% of appealed district courtsentences under the Penal Code were revised by the Courts of Appeal. His study does not, however, attempt to examine the Appeal Courts' reasons for modification.

12 Minnesota's grid, and these aggravating and mitigating circumstances, are examined in von Hirsch, Knapp, and Tonry, chap. 5. 

Principles for choice of sentence 5stances. The decision whether to remain within the grid range or to deviate is a matter of judgment—around which a considerable body of case law has been accumulating in Minnesota.13
    The grid has, however, another disadvantage: the numbers tend to overshadow the underlying principles. Minnesota's guideline-writing agency purportedly chose a rationale emphasizing proportionality similar to that adopted by Sweden's Imprisonment Committee. Under that rationale, the seriousness of the current offense would be the primary determinant of the sentence—and the criminal record would have only a secondary role. When the drafters of Minnesota's guidelines started filling in the numbers on the grid, however, they decided to impose prison sentences on lesser felons having long criminal records. This was done to enhance the political acceptability of the guidelines, but it was inconsistent with the guidelines' rationale and eventually generated a host of practical problems as well.14 Deciding on the numbers took on a life of its own that made principled choices more difficult to implement.
    The attraction of the Swedish Imprisonment Committee's proposals on choice of sentence is that they consist of principles. The primary factors to be considered in deciding the sentence are set forth, not the actual sentencing outcomes. The proposed statute thus provides that the seriousness of the crime should be given principal emphasis, directs how seriousness (straffvärde) should be judged, and gives broad, general directives on the use of imprisonment and of lesser sanctions. The numbers—the actual quanta of sentences—are to be evolved later by the courts. This allows the statute to focus on what is important: the policy. Consider Minnesota's problem of what should be done with offenders who are convicted of crimes of lesser penal value, but who have substantial criminal records. The ImprisonmentCommittee's proposals offer a general policy: since the seriousness ofthe crime should count most, such offenders ordinarily should not receive the severe sanction of imprisonment. Such a general statute need not address the extraordinary cases: what should happen if the offenders criminal record is extremely long—for instance, the case of the person convicted of a routine larceny for the twentieth time. The courts are competent to deal with such extraordinary cases, bearing the statute's general principles in mind. The drafters did not need to distort the general principles to supply a politically "acceptable" solution for the special cases.


13 The cases are summarized in Knapp, Minnesota Sentencing Guidelines and Commentary.

14 von Hirsch, Knapp, and Tonry, chaps. 3, 5 och 8.


6 Andrew von HirschIII. Sources and structure
The predecessor of the Imprisonment Committee's proposals is the Finnish sentencing statute. In 1976, Finland added a new Chapter 6 to its Penal Code, dealing with choice of sentence. The most important provision of that chapter, its first section, makes the seriousness ofthe crime of conviction the chief determinant of severity of sentence. It provides that the sentence should be "in just proportion" to (1) the harm caused or risked by the offense, and (2) the guilt of the offenderas manifested in the offense.15 These same two elements, "harm" and "guilt" (skada and skuld) are the defining elements of a crime's penal value (straffvärde) under the Imprisonment Committee's draft. Finland has now had a decade of experience with its law, and sentencing patterns appear to be changing toward giving increased attention tothe degree of seriousness of the criminal conduct.16 The provisions also seem to have developed a substantial consensus of support, for virtually no sentiment favouring repeal or major alternation is discernible in Finland to date.
    The Finnish statute, however, leaves some major questions unaddressed. One question concerns the use of imprisonment. Should a large variety of crimes be punished by imprisonment, albeit with graded durations reflecting the comparative seriousness of crimes? Or should imprisonment be treated as a last resort, appropriate only forcrimes that are quite serious? The statute does not say. Another issueis the role of the previous criminal record. The statute makes having arecord of previous convictions an aggravating factor, if the record suggests "the offender is apparently heedless of the prohibitions of law."17 This language leaves uncertain, however, how much weight should be given the criminal record.
    The proposed Swedish provisions would give the courts better guidance, because the steps in determining the sentence are more clearly indicated. First, the penal value of the crime-category is to be determined, based on the applicable penalty scales and on the harmfulness and culpability that is typical for such conduct. Next, the penal value of the offender's particular criminal act is ascertained, considering also the presence or absence of aggravating and mitigating factors—and these are spelled out more fully than they are in the Finnish statute.18 Next, comes the choice of type of sanction, which


15 Penal Code of Finland 6:1.

16 This is based on conversations with Tapio Lappi-Seppälä, who is now conducting a study of the Finnish courts' implementation of this statute.

17 Penal Code of Finland 6:2 4.

18 Proposed 33: 3 and 33: 5 BrB.


Principles for choice of sentence 7depends on whether the penal value thus ascertained is high, medium, or low. With certain exceptions to be discussed later, imprisonment is to be reserved for offenses with high penal value.19 The role of the prior criminal record is addressed, as we shall see. Treatment and predicted likelihood of recidivism may be considered only in certain specified circumstances, which the statute spells out.20
    Nevertheless, the proposed law constitutes a body of principles, not a table of numbers. The law spells out the penal aims to be achieved, and offers a broad framework for deciding when imprisonment would and would not be the appropriate sanction. The tariff—that is, the pattern of actual sanctions—would be evolved by the courts over time. The difference between the Imprisonment Committee's proposed approach and the current situation is that the courts would no longer be working in a vacuum. They would receive guidance as to aims and the general shape of the sentencing structure, and the steps to be considered when a sentence is imposed. This would put them in a better position to decide how much punishment various kinds of crime should receive.
    Consider, for example, the crime of residential burglary, which seems to be generating increasing concern in the larger Swedish cities. At present, the courts have almost no useful guidance. What sentence might rehabilitate most burglars is unknown. Trying to distinguish potential recidivist from other burglars is likewise problematic, inview of the ethical and evidentiary difficulties involved.21 And so forth. Given the confusion of aims and the competing potential criteria, it is not surprising that the courts have difficulty fashioning a consistent response. Were the Imprisonment Committee's proposed chapters on choice of sentence to become law, matters would become more manageable. The sentencing court would no longer need toselect the purpose to be achieved: it would not have to resolve whether it was seeking to deter burglars, incapacitate them, treat them, or whatever. The statute would already have established the predominant aim: imposition of a proportionate sanction. The law would also supply the criterion for proportionality: namely, the penal value of the offense. The court's initial job, therefore, would be to assess burglary'spenal value. How much harm does burglary do or threaten? How culpable is the conduct? What grounds of mitigation might various kinds of burglars have? Deciding this is conceededly no easy task, but the courts and the parties would at least have the issues framed for


19 Proposed 34:6.

20 See discussion below.

21 These problems are spelled out in Past or Future Crimes, chs. 9-12.


8 Andrew von Hirschthem. There would be an incentive to begin to break burglaries down into different types and to distinguish among their respective penal values. Since only certain subspecies of burglary would be likely to berated high in penal value, only those would ordinarily receive a prison sentence. The courts would retain their primary responsibility for deciding sentence severity, but now would have assistance in reasoning through such decisions. The lawmaker would do what the drafters of a statute are well positioned to do: establish aims and policies for the system as a whole which the courts would then implement in greater specificity.


IV. Rationale
Why the emphasis on proportionality? As I have written elsewhere at some length on the rationale of proportionality,22 let me just summarize. In order to support this idea, we need no deep, "metaphysical" theories of requital for evil, or of guilt and atonement. The rationale is much simpler and more straightforward.
    The various benefits and burdens distributed among a state's citizens fall into two main types. The first are those that are neutral on their face-wages, social-welfare payments, taxes, etc. Other than being agreeable or disagreeable, these do not carry any clues about how they should be allocated: there is nothing about money per se, that tells one whether it should be taken or given according to merit, need, or whatever. Hence the continuing philosophical and political debate about how such things might fairly be distributed. The second type are benefits or burdens that have approbation or censure attached to them as part of their defining characteristics. The prize is an example. A prize is a thing of value (or its symbolic representation) that connotes approval for its recipient for his or her conduct. Punishmentis the reverse. It is a deprivation visited on someone under circumstances of censure or klander. Other responses run the gamut from praise to blame, such as university grades—where a high grade connotes approval and a low grade its opposite.
    When things thus connoting praise or blame are distributed, it should be according to the desert of the person—that is, according to the merits of his or her performance. If a rich student and a poor student do equally good work, the university might wish to give the indigent student extra financial support to help him afford to pursue his studies. If the money is designated as an award, however, only the comparative merit of those students' performance should count. Simi-


22 Past or Future Crimes, chaps. 3—5; von Hirsch, Nyklassicism, proportionalitet och straffets grunder. 

Principles for choice of sentence 9larly, in a contest it is inappropriate to say that A won but B should get the prize because he or she needs the money more or would put it to better use.
    The same logic holds for punishment—and undergirds the idea of proportionality. Punishment involves censure as part of its definition. The only difference between a tax and a fine lies in the condemnation involved in the fine. Yet that added element of censure is critically important. Sweden is a country of high taxes and comparatively modest fines—yet almost everybody would prefer paying their taxes to paying a fine, because of the blame the latter involves. Were the critical element of censure removed, the criminal sanction would become unrecognizable. This is brilliantly evoked in Samuel Butler's novel Erehwon, published in England a century ago. In Butler's strange utopia, people are punished for being ill but treated forcriminal acts. They thus feel deeply ashamed of succumbing to bronchitis, but discuss their therapies for having committed fraud whith nosense of shame. Since those therapies are painful, some people still tryto avoid committing crimes. But the topsy-turviness of such a world is made all too apparent.
    It is because punishment involves censure that its principal criterion should be the blameworthiness—that is, the degree of seriousness—of the criminal conduct. If A is punished more than B, this connotes that A's conduct is disapproved of more than B's, and that is appropriate only if his conduct was, indeed, worse. Why do we feel disparity of sentence to be offensive? It simply is that if A's and B's criminal conduct is equally reprehensible, then giving them unequal punishment implies, inappropriately, that one has behaved more reprehensibly than the other. Departing from the requirements of proportionality is unjust, in other words, not because it fails to requite suffering with suffering—but because the offender is being treated as more or less worthy of censure than the harmfulness and culpability of his conduct warrants.
    The Imprisonment Committee's proposals do not eliminate ideas of general prevention, but give such ideas a different (and in my judgment, more workable) role. Rather than being used to decide particular penalties, general prevention serves as a support for the system of punishments as a whole. Thus:
    1. General prevention helps justify the existence of criminal sanctions. Conduct is criminalized so that persons will desist from it. Imposing a sanction—an unpleasant consequence—is threatened so that persons will take the prohibitions seriously.23 Having the sanc-


23 See Jareborg, Straffets syften och berättigande; Past or Future Crimes, chap. 5.


10 Andrew von Hirschtion involve censure enhances legal prohibitions' influence,24 as well as reflecting moral judgments about the blameworthiness of victimizing conduct.25
    General prevention (or our beliefs about it) may also affect the overall lenience or severity of the penalty system. Proportionality is primarily a doctrine about the comparative severity of penalties. It requires that equally reprehensible criminal conduct should be punished equally, and that sanctions should be graded in relative severity to reflect the seriousness of crimes. It does not—and is not supposed to—provide precise starting or anchoring points. When penalties have been graded to reflect the seriousness of crimes, one still may make pro rata increases or decreases in the penalties (within reasonable limits) without disturbing proportionality.26 This could allow room to consider general prevention, were we to develop the necessary empirical knowledge, in deciding overall severity levels.27
    2. Where general prevention ordinarily is not appropriate is for resolving specific sentencing questions: for deciding whether Offense A should be punished more or less onerously than Offense B. Using general prevention to decide such questions involves two kinds of difficulties. The first is evidentiary. Criminalogists have done considerable research on a major subspecies of general prevention—namely, deterrence. To determine sentences on the basis of their deterrent effects, one needs information on how the rates of various crimes are affected by changes in penalties. Little information of that sort has been obtainable. Although there is reason to believe that penalizing conduct deters better than having no penalty would, scant capacity exists for measuring the magnitude of deterrent effects with respect to particular species of crime—because crime rates are affected not only by penalty levels but by so many other economic and social factors.28 The second objection is ethical. A deterrence calculus would require penalties to be judged not by the injuriousness and culpability of offenders' criminal choices, but by contingent facts about the responsiveness of potential perpetrators to threatened penalties. If Offense A is less serious than Offense B, it hardly seems fair to give it enhanced punishment, because its potential perpetrators can be intimidated,


24 This theme is emphasized in the BRÅ report, Nytt Straffsystem.

25 This theme is emphasized in Past or Future Crimes, chap. 5, and in von Hirsch, Nyklassicism, proportionalitet och straffets grunder.

26 This distinction between comparative proportionality and the fixing of anchoring points is elaborated in Past or Future Crimes, chaps. 4 and 8.

27 For how that might be done, see von Hirsch, Doing Justice, chap. 16.

28 See the report of the National Academy of Sciences panel on deterrence research, set forth in Blumstein, Cohen and Nagin. 

Principles for choice of sentence 11wheras potential perpetrators of Offense B seem unresponsive to threats.29


V. Penal value
How, then, should proportionate sentences be determined? The proposal's central concept is that of a crime's penal value, and the draft begins with a general definition (33: 2): "The penal value (straffvärde) of a crime is determined by its seriousness (brottets svårhet)." Seriousness involves the concepts discussed above: of the degree of reprehensibleness or blameworthiness (klandervärdhet) of the conduct.
    To determine a crime's seriousness, the proposal goes on to state, special regard should be given to (1) the harmfulness of the conduct, and (2) the personal culpability of the actor (33: 2). This definition—of seriousness in terms of the conduct's harm and culpability—is standard in the recent literature on proportionality.30
    Harm. That harm is an important element in a crime's seriousness should be obvious. Murder is more serious than assault because the harm characteristic of such conduct is greater: death instead of injuryor attemped injury. Harm has always been important in determining the statutory penalty scales—which is why the legal maximum and minimum for murder are so much higher than those for assault. What the proposal would do is to direct the judge to give more careful consideration to the conduct's harmfulness within the applicable scale: that is, to try to distinguish among types of assaults in terms of the degree of actual or potential injuriousness.


29 The above account thus involves (1) a justification for criminalizing conduct that is, at least in part, preventive and utilitarian, but (2) a theory of deciding sentences—that is, of allocating punishments among convicted offenders—that is non-utilitarian and based on notions of proportionality and desert. There is no contradiction in having such differing explanations for the existence and the distribution of punishments—as the Oxford legal philosopher H. L. A. Hart pointed out nearly three decades ago in his well known essay, Prolegomenon to the Principles of Punishment. (The essay is reprinted in Hart, chap. 1.) Why this distinction? On the Committee's theory, the criminal sanction's existence is justified on preventive grounds, as discouraging certain undesired behavior. It does soin part because of its inhibition-strengthening features. Through its censuring implications, punishment helps reinforce people's sense of the wrongfulness of criminal behavior. Once punishment is established, however, it should in fairness be distributed among convicted offenders consistently with its censuring implications. The Committee's argument is thus similar to one which might naturally be used to explain academic grades. If one askes why a school or university should grade students for their work, the answer would naturally be utilitarian and forward looking: grades are an incentive for better work. But if one asks why, once such a system has been established, a good student should be entitled to a good grade, the answer should be retrospective and desert-oriented. Since high grades symbolize approval for good performance, it would not be fair to deny such a grade to someone who has done fine work, or to give such agrade to someone whose work has been poor. See, Past or Future Crimes, chap. 5.

30 See, e. g., ibid., chap. 6.


12 Andrew von Hirsch    Giving harm this central role should stimulate the development of more sophisticated doctrines on how to assess harm. A simple criterion would be that of violence: crimes are to be deemed more injurious, as the extent of physical injury they visit or threaten increases. But this is not wholly satisfactory: one can think of a variety of offenses, such as major economic crimes, that appear quite serious but nevertheless are not crimes of violence at all. Recently, I have suggested a broader standard: harms may be graded according to the degree to which they characteristically restrict people's ability to direct the course of their own lives.31 The gravest harms are those which interfere with almost any choice a person might wish to make. This accounts for our sense of the gravity of violence, for violence restricts victims' choices so drastically. (The person who is murdered has no choices left at all, the person who is seriously injured has his remaining choices sharply curtailed.) The theory also accounts for the harmfulness of certain non-violent crimes. A person cannot order his life if deprived of his means of economic subsistence. Economic crimes that typically destroy or threaten people's means of livelihood (for example, by swindling them of their savings) involve grave harm. Such a conception cannot be applied automatically, and needs to be supplemented by theory for dealing with crimes primarily injurious to collective interests, such as corruption or tax evasion. But it is, perhaps, a start.
    Culpability. It is a peculiarity of the Swedish language, like the German, that the word "culpability" does not exist. Hence the drafters had to resort to the more old-fashioned term, guilt or skuld (the German equivalent is "Schuld"). While the word "guilt" may for some readers evoke theological connotations, those are not intended. When the draft speaks of "the offender's guilt manifested in the conduct" (33:2 (2)), it is not referring to an elusive evil state in the criminal's soul which the conduct reflects. The draft is referring, instead, to what in English would be referred to by the more neutral word culpability: the degree of dolus or culpa of the conduct, the presence of partial excuses, and so forth.
    Existing Swedish substantive law can thus provide guidance interpreting the culpability concept. Consider the role of dolus and culpa. The law relies already upon the degree of intentionality, recklessness, or negligence of the conduct in fixing the penalty scales. The difficulty has been that, within the applicable scales, the courts have not had the incentive to accord such distinctions much weight. By


31 For a fuller account, ibid., chap. 6, and Feinberg, 37—45, 185—214.


Principles for choice of sentence 13explicitly emphasizing culpability, the draft would call upon the courts to calibrate the individual sentence according to the degree of intentionality or negligence involved.
    Swedish substantive law has also had elaborate doctrines of excuse32—but the sentencing analogue, of partial excuse, has not been given much attention. The draft's provisions on culpability would require this attention to be given. Here, the courts will also be given assistance by the draft's special provisions on mitigating circumstances. Cases where the actor's mental condition makes him less than normally culpable are explicitly addressed (33:5 (2) and 33:5 (3). Provocation is extended beyond the law of homicide, to become a possible mitigating circumstance for any charge (33:5 (1)).33


VI. General criteria for imprisonment
The proposal sets forth, for the first time in Swedish law, criteria for imprisonment.34 These criteria, found in proposed 34:6 BrB, indicate when imprisonment ordinarily is to be imposed, in preference to the lesser sanctions of probation or conditional sentence. The criteria when read together with the draft's other provisions-prescribe imprisonment in two main kinds of cases. The first is where the crime of conviction is serious: in the words of the draft, where it has "considerable penal value" (betydande straffvärde). The courts will have to determine which crimes thus qualify as serious, but I expect that crimes of violence such as armed robbery would be included, as would major economic offenses. Those convicted of such crimes could expect to be imprisoned, unless they were able to establish mitigating circumstances indicating reduced culpability for the crime itself. The fact that the defendant was a first offender ordinarily would not justify with holding imprisonment. The rationale is plain enough. If punishment is to reflect the gravity of the conduct, then the system's most severe type of sanction, imprisonment, becomes appropriate for the worst conduct.
    The second type of case where imprisonment would be invoked concerns the criminal record. The Committee's proposals generally attempt to restrict the role of the prior record: serious offenders wouldbe confined even if not previously convicted, and lesser offenders are to be given non-prison sanctions even if recidivists. However, the prior


32 These doctrines are summarized in Jareborg, Uppsåt och oaktsamhet.

33 For a discussion of provocation and its rationale, von Hirsch och Jareborg, Provokation.

34 The draft also sets forth parallel criteria regarding the choice between fines vs. conditional sentence or probation. 

14 Andrew von Hirschrecord would continue to play a role with respect to offenses in the upper-middle range of seriousness: those not quite grave enough to warrant the sanction of incarceration in themselves, but nearly so. (An example might be burglary). Offenders convicted of such crimes would initially receive an intermediate punishment — probation or conditional sentence, coupled perhaps with a fine (see 34:11). Once such an offender has accumulated a significant criminal record, however, incarceration would become the sanction of choice.
    Why treat the record in this fashion? Giving the criminal record limited weight can be justified, I believe, in a system based on proportionality and desert. Treating being a first offender as an extenuating circumstance is a way of recognizing human fallibility in our criteria of punishment. By being given a somewhat scaled-down punishment, the first offender is censured for his act—but given some respect for thefact that his inhibitions against wrongdoing have functioned on previous occasions, and accorded some sympathy for the all-too-human frailty that can lead someone to such a lapse. With repetition, however, this mitigation is progressively lost, and the person is held fully accountable.35
    A desert-based rationale, however, cannot give the prior record primary weight, as a predictive sentencing scheme would. (To the extent criminality can be predicted at all, it is so chiefly on the basis of how early and often the offender previously has offended36.) The reason it cannot is that such a philosophy assumes—as we saw above—that punishment plays a critical role in expressing valuations of conduct. Giving public acknowledgement that some kinds of crimes are more reprehensible than others is an important part of what the criminal sanction is about. A sentencing system based chiefly on the criminal record would make that public valuation much more diffuse. The punishment would cease to express the degree of heinousness of what was done—and shift the attention, instead to who did the act and what he did before.37


VII. Special provisions for other penal aims
While the Committee's scheme emphasizes penal value and proportionality, it is not intended as a single-aim proposal. Deterrent, incapacitative and rehabilitative aims have their roles—only those roles are more carefully defined than in existing law.


35 For fuller explanation, see Past or Future Crimes, chap. 7.

36 Ibid., chap. 11.

37 Ibid., chap 7.


Principles for choice of sentence 151. Deterrent Sanctions: The "Special Nature" Provision
Besides the general criteria for imprisonment just discussed, the draft contains an additional ground for imprisonment: that "special reasons" exist for confinement with regard to "the nature of the crime"(34:6). On its face, this provision may seem puzzling. Since the draft has expressly addressed the crime's penal value and the offender's criminal record as grounds for imprisonment, what "special nature" is being spoken of? The Committee's report provides the answer. This provision is designed to authorize use of imprisonment for cases such as drinking and driving, where imposition of more than usually severe punishments is the country's considered policy. The strategy, here, is one of deterrence. A prison sentence is to be imposed for drinking and driving (not withstanding that the defendant's personal culpability may not be high) in order to get drivers to desist from this hazardous activity. But such sanctions can only be invoked, the Committee's report indicates, when Parliament has given its specific endorsement.38
    In adopting such a proposal, the Committee evidently is treating proportionality as a presumption of fairness, but not an absolute rule. The presumption ought ordinarily to be observed, but can be over ridden if the counter-considerations are urgent enough. Crime prevention could constitute such an overriding reason but only when the social harm to be prevented is very grave. Judges would not be permitted to base sentences on deterrence—because they generally lack the information to gauge the preventive effects of such penalties, and do not have the necessary general policy-making authority. Requiring special parliamentary approval is designed to assure that (1) sound empirical reasons exist for assuming a policy of deterrence might be effective, and (2) the conduct is so severely socially harmfulas to warrant departure from proportionality requirements.
    Only time will tell how sound this provision is. It assumes that Parliament will act with considerable self-restraint: that it will authorize other deterrent punishments only when the evil involved is of a gravity comparable to drinking and driving, and the evidence favouring the strategy of imprisonment is comparably persuasive. Should the evidentiary standards deteriorate, or should the provision be used to take posturing anti-crime stances, it could undermine much of therest of the proposed statute. I would hate to see an American state legislature being called upon to make such determinations, but one hopes the Swedish Parliament will be more responsible.


38 SOU 1986:14, pp. 457—58. Such parliamentary endorsement may be set forth in preparatory works rather than in an actual statute. 

16 Andrew von Hirsch2. Incapacitative Sanctions
To what extent should the sentence be based on the offender's likelihood of returning to crime? In the sentencing literature of the past two decades, two major objections have been made to such a practice. The first is that the capacity to predict recidivism accurately is limited, and to predict on the basis of the restricted information available to sentencing courts more limited, still.39
    The second objection is that predictive sentencing would constitute punishment for crimes not yet committed. This latter objection can be stated a little more precisely. In so far as any capacity to predictrecidivism exists, the factors on which one must rely are quite different from the crime's seriousness: primarily, they involve the offender's early criminal record and his social and employment history. Predictive sentencing, therefore, will violate the principles of proportionality because sentence severity will have to depend so much on factors alien to the gravity of the offenses.40 The Committee, by and large, shares this scepticism—and does not authorize judges to choose sentence on the basis of the supposed risk posed by the individual defendant.
    The Committee provides an interesting qualification, however. Predictive sentencing raises problems of proportionality when the sentencer proposes to punish persons who supposedly are high risks substantially more severely than those who are low risks. The objection diminishes, however, when it is used to change the character of the sentence, but not its severity by much. If high risk offenders get Type A punishments and low risk offenders get Type B, proportionality is not offended when the two types of punishment are of approximately equivalent severity. Therefore, the Committee included a provision (34:10) to the effect that the sentencing judge, when choosing between conditional sentence and probation, may choose probation when he or she finds that probation supervision would help induce the offender remain law-abiding. Probation-with its conditions and supervision—may be somewhat more onerous than conditional sentence, but as traditionally conducted in Sweden, is not much more so. If the severity-difference between the two sanctions is not great, deciding between them on grounds of risk is no great affront against proportionality. The Committee thus allows scope for special preventive aims, but not in a manner that would infringe the fundamental goal of fair, proportionate sanctions.


39 See Past or Future Crimes, chaps. 9 and 10.

40 Ibid., chap. 11.


Principles for choice of sentence 173. Rehabilitation
The Committee's report is sceptical about the effectiveness of penal rehabilitation strategies, and I think rightly so. Research on penal treatments has suggested that we still do not know much about which treatments successfully inhibit further criminality.41 However, the draft is concerned only with the choice of sentence—and particularly with how that choice affects the sentence's severity. Aside from the question of effectiveness, rehabilitation raises problems of justice when used to decide severity of sentence. Suppose A and B commit a burglary together. If A goes to prison because he is deemed "unresponsive" to treatment and B is put on probation because he is considered amenable to a community-based rehabilitation program, this infringes proportionality: the two offenders, having committed an equally reprehensible act, receive punishments of substantially differing onerousness because of personal characteristics they happen to have. A system which takes proportionality seriously should have a presumption against this kind of sentencing.42
    Where differences in severity are not involved, however, proportionality is not at issue. Here, the draft would permit the use of penal treatment. Prison authorities could continue to offer rehabilitative programs for offenders committed to prison; probation authorities could continue to provide programs and services to offenders on probation, and the provision just spoken of (34:10) would allow rehabilitative as well as predictive considerations to influence the choice between probation and conditional sentence.43
    The draft also authorizes the rehabilitative sentence in one special case. The court can reduce a sentence from imprisonment to probation or conditional sentence where "through the offender's own efforts, a considerable improvement has occurred in his personal and social situation that bear on his criminality" (34: 7 (2)). The provision's reference to change wrought by the offender himself is an attempt to restrict the treatment-based sentence to its most plausible and sympathetic type of case. I happen to doubt the value of the exception: it involves a sacrifice of proportionality without countervailing reasons that I would consider sufficiently persuasive. But it does indicate the Committee's intention to create a multi-purposesystem in which rehabilitation would continue play a modest role.


41 See, for example, the National Academy of Science report on rehabilitation research, set forth in Sechrest, White and Brown. For recent Scandinavian research, see Kühlhorn.

42 For the ethical problems of such sentences, see von Hirsch, Doing Justice, chap. 15; for their problems of effectiveness, see Kühlhorn.

43 For the greater potential for effectiveness of such treatment strategies, see Kühlhorn. 2-37-161 Svensk Juristtidning 

18 Andrew von HirschVIII. Conclusions
The Imprisonment Committee's choice-of-sanction proposals strike me as providing workable criteria for deciding how much punishment convicted criminals should receive. By emphasizing proportionality, they promote fairness—while avoiding unrealistic expectations about the preventive impact of particular sentencing decisions. By setting forth general criteria, they offer a coherent policy for sentencing—while leaving its implementation in a tariff of sentences to development by the courts.
    The Committee's proposals presuppose courts that are responsive to policy directives stated in law. Swedish courts have had a history of that kind of responsiveness. A Swedish judge, much more than his orher American counterpart,44 can be expected to take broadly-statedlegal criteria seriously, and to familiarize him- or herself with the preparatory works in applying them. The proposals, in short, assume a judiciary that is willing to be guided by a light rein, and that does not need a table of numbers in a sentencing grid to guide its practice. Sweden has such a judiciary, and the Committee's proposals, if enacted, should help bring more order, purpose and equity to sentencing decisions.


Ashworth, Andrew. Techniques of Guidance on Sentencing. Criminal Law Review 1984 pp. 519 ff.
Blumstein, Alfred, Jacqueline Cohen and Daniel Nagin, eds. Deterrence and Incapacitation: Estimating the Effects of Criminal Sanctions on Crime Rates. Washington, D. C. 1978.
Feinberg, Joel. Harm to Others. New York 1984.
Hart, H. L. A. Punishment and Responsibility. New York 1968.
von Hirsch, Andrew. Deservedness and Dangerousness in Sentencing Policy. Criminal Law Review 1986 pp. 79 ff.
von Hirsch, Andrew. Doing Justice: The Choice of Punishments. New York 1976. Reprint ed.: Boston 1986.
von Hirsch, Andrew. Guidance by Numbers or Words? - Numerical vs. Narrative Guidelines for Sentencing. In K. Pease and M. Wasik, eds. Sentencing Reform: Guidance or Guidelines? Manchester 1987.


44 For the problems of broad legal principles of sentencing in an American context, see von Hirsch, Guidance by Numbers or Words? 

Principles for choice of sentence 19von Hirsch, Andrew. Nyklassicism, proportionalitet och straffets grunder. NTfK 1982 pp. 97 ff.
von Hirsch, Andrew. Past or Future Crimes: Deservedness and Dangerousness in the Sentencing of Criminals. Manchester 1986.
von Hirsch, Andrew och Nils Jareborg. Provokation. NTfK 1986 pp. 436 ff.
von Hirsch, Andrew och Nils Jareborg. Påföljdsbestämning i USA. BRÅ Rapport 1984:4.
von Hirsch, Andrew, Kay A. Knapp and Michael H. Tonry. The Sentencing Commission and Its Guidelines. Boston 1987.
Jareborg, Nils. Straffets syften och berättigande. TfR 1985 pp. 1 ff.
Jareborg, Nils. Uppsåt och oaktsamhet. Uppsala 1986.
Knapp, Kay A. Minnesota Guidelines and Commentary. St. Paul, Minn. 1985.
Kühlhorn, Eckart. Allmänprevention och individualprevention. SOU 1986:15, bilaga 2. pp.27 ff.
Löfmarck, Madeleine. Ett lagom ideologiskt betänkande? NTfK 1986 pp. 407 ff.
Nytt Straffsystem. BRÅ Rapport 1977:7.
Past or Future Crimes. See von Hirsch, Past or Future Crimes.
Sechrest, Lee, Susan O. White, and Elizabeth Brown, eds. The Rehabilitation of Criminal Offenders: Problems and Prospects. Washington, D. C. 1978.
Thomas, D. A. Principles of Sentencing. London 1979.