Volume 29 - Issue 1

On Barth’s Denial of Universalism

By Oliver D. Crisp

Oliver Crisp teaches theology at St. Mary’s College, University of St. Andrews. He is a regular reviewer for Themelios.

It is notorious among theologians that Karl Barth defends doctrines of election and atonement that appear to lead to universalism, but that Barth steadfastly maintained did not lead to universalism. As Jüngel records it, Barth emphatically claimed, ‘I do not teach it (universalism), but I also do not teach it’.1 There have been many who are willing to defend Barth in this matter. For instance, John Colwell claims that:

[I]f some of Barth’s critics refuse to take this divine freedom seriously with respect (especially) to Barth’s doctrine of election and consequently suspect him of implicit universalism then that is their problem rather than his and probably says more about them than it says about him.2

Similar views can be found among other barthians both present and past. Thus, for example, Joseph Bettis says that ‘for Barth, one can reject both Arminianism and double predestination without having to accept universalism’.3

There, however, have also been those voices raised in opposition to this view. For instance, Hans Urs von Balthasar says:

It is clear from Barth’s presentation of the doctrine of election that universal salvation is not only possible but inevitable. The only definitive reality is grace, and any condemnatory judgement has to be merely provisional.4

Similarly, G.C. Berkouwer observes that:

In original universalism, the issue is a universal offer because Christ died for all, and election remains in the background for the moment. But with Barth, Christ’s death touches precisely upon the election of all, which election has become manifest in Christ’s death. The universality of the message is no longer at odds with the fact of election, for it is based on the universality of election.5

Even Geoffrey Bromiley, one of the translators of Barth’s Church Dogmatics into English, and a theologian sympathetic to Barth’s account of election, nevertheless closes his own overview of Barth’s doctrine of election with these words: ‘The ambivalence at this decisive point—will all be saved or not, and if not, why not?—by no means outweighs the solid merits of Barth’s presentation.’6

So what is the logic of Barth’s position? Does his view yield a version of universalism or not? In this essay, I will attempt to show that Barth’s doctrines of election and atonement do indeed yield a version of universalism, despite the protestations of both Barth and his defenders to the contrary.

Barth’s doctrine of atonement and election

The argument depends upon a number of theological terms pertaining to universalism which it might be helpful to explain at the outset. First, I take it that the term ‘universalism’ refers to a family of similar views which share in common the notion that all humanity will be saved by God. None will be finally damned to hell. In the current literature these different versions of universalism have been categorised into two groups: necessary and contingent universalism. Necessary universalism is the view that it is not just true, but necessarily true that all humanity will end up in heaven. By contrast contingent universalism states that, although a human being could be consigned to hell, as a matter of contingent fact no-one will end up there.7

Second, I will use the terms ‘elect’ and ‘reprobate’ to refer to two groups of humanity. The former term denotes that group which God decrees to save; the latter, that group which God decrees to damn.

With these clarifications in mind, let us turn to Barth himself. I take it that Barth’s denial of universalism depends upon his doctrines of election and atonement in particular. In the argument that follows, I will show that Barth’s denial of universalism is either disingenuous (he was a universalist), or just plain muddled (his position is not coherent). There is a third option: that Barth did not see the logical implications of his position. However, this seems unlikely, so I will ignore it. Either result means Barth’s denial is false.

The argument depends on two assumptions:

A1. There is a domain of moral agents comprising all human agents.

I will not comment on the fate of angelic moral agents, if there are any. Nor is the subject of angelic moral agents one which Barth touches upon in any great detail with respect to his doctrine of election. In common with the Augustinian tradition, Barth appears to believe that the question of the fate of demonic angelic agents is a separate issue from that of human moral agents.

A2. By Christ’s death the sin and guilt of those for whom he died is atoned for.

Barth clearly does believe this. However, I will not discuss the mechanism by which Christ’s atonement is achieved according to Barth. All the argument requires is that his death brings about the atonement for sin required for those who are elect-in-Christ.

First, I will set out Barth’s argument for atonement. This can be found in passages from the Church Dogmatics like the following:

The rejection which all men incurred, the wrath under which all men lie, the death which all men must die, God in his love for men transfers from all eternity to him in whom he loves and elects them, and whom he elects at their head and in their place.8


For if God Himself became man, this man [Christ], what else can this mean but that He declared Himself guilty of the contradiction against Himself in which man was involved; that He submitted Himself to the law of creation by which such a contradiction could be accompanied only by loss and destruction; that He made Himself the object of the wrath and judgment to which man had brought Himself; that he took upon Himself the rejection which man had deserved; that He tasted Himself the damnation, death and hell which ought to have been the portion of fallen man?… If we would know what it was that God elected for Himself when He elected fellowship with man, then we can answer only that He elected our rejection. He made it his own. He bore and suffered it with all its most bitter consequences.9

We can express his argument for the atonement of Christ in the following way:

  1. Given A1 and A2, Christ’s death atones for the sin of all human agents.

By this Barth seems to mean that:

  1. Christ’s death is sufficient and efficient for all human agents.

That is, Christ’s death is not simply potentially universal in scope (it could save all humanity); it is actuallyuniversal in scope (all humanity are saved by it). It might be argued that Barth’s position is merely that the atonement is universal in scope, not effectiveness. However, that Barth’s position does involve a universally efficient atonement can be seen from passages such as the following:

There is no-one who does not participate in Him [Christ] in His turning to God. There is no-one who is not … engaged in this turning. There is no-one who is not raised and exalted with Him to true humanity. ‘Jesus Christ lives, and I with Him.’10

We shall return to this issue at a later point. For the present, let us assume that Barth does endorse something like 2 above. From here we move to:

  1. This work is completed at the cross.
  2. This work is appropriated not via the traditional Reformation formula: ‘If you repent and believe, you will be saved; if you do not repent and believe, you will not be saved’, but by agents coming to realise that, ‘this is what God in Jesus Christ has done for your sake. Therefore repent and believe,’11

This raises a question. It is this: if the atonement is universally effective, according to Barth, then how does this tie into his doctrine of election? To answer this, let us lay out Barth’s doctrine of election in brief. This aspect of Barth’s theology can be found in passages such as the following:

This, then, is the message with which the elect community (as the circumference of the elect man, Jesus of Nazareth) has to approach every man—the promise, that he, too, is an elect man. It is fully aware of his perverted choice. It is fully aware of his godlessness … It is fully aware, too, of the eternal condemnation of the man who is isolated over against God, which is unfailingly exhibited by the godlessness of every such man … It knows of the wrath and judgment and punishment of God in which the rejection of the man isolated over and against God takes its course … It knows that God, by the decree He made in the beginning of all His works and ways, has taken upon Himself the rejection merited by the man isolated in relation to Him; and on the basis of this decree of His the only truly rejected man is His own Son; that God’s rejection has taken its course and been fulfilled and reached its goal, with all that that involves, against this One, so that it can no longer fall on other men or be their concern. The concern of other men is still the sin and guilt of their godlessness—and it is serious and severe enough. Their concern is still the suffering of the existence which they have prepared for themselves by their godlessness (in the shadow of that which the One has suffered for them)—and it is bitter enough to have to suffer this existence. Their concern is still to be aware of the threat of their rejection. But it cannot now be their concern to suffer the execution of this threat, to suffer the eternal damnation which their godlessness deserves. Their desire and their undertaking are pointless in so far as their only end can be to make them rejected. And this is the very goal which the godless cannot reach, because it has already been taken away by the eternally decreed offering of the Son of God to suffer in place of the godless, and cannot any longer be their goal.12

We can express what Barth says about election thus:

  1. Christ is the Elect One. (That is, the set of the elect comprises one member, Christ.)
  2. Christ is the Reprobate One. (That is, the set of the reprobate comprises one member, Christ.)

In this way, Barth’s doctrine fuses the so-called ‘double decree’ of Calvinism in the person of Christ, who is both the Elect and Reprobate One. But the way in which this is applied to the set of human agents is asymmetrical.

  1. All human agents are elect only in the derivative sense of having a saving relation to the set of the elect and its single member, Christ.


  1. The sin of all human agents is atoned for by Christ, the Reprobate One, who is the only member of the set of the reprobate.

Rather than 8, it might be tempting to construe Barth as saying something more like:

8*. All human agents are reprobate only in the derivative sense of having a relation to the set of the reprobate and its single member, Christ.

But this would entail:

  1. All human agents are simultaneously members of the sets ‘elect-in-Christ’ and ‘reprobate-in-Christ’.

This which is incoherent, for then, all humanity would be derivatively elect and reprobate simultaneously, and, presumably, co-terminously. This would, of course, be absurd. (Compare the idea that someone could simultaneously be both an associate member and non-associate member of a country club.) I suggest that Barth must mean 8 rather than 8*, in order for his argument to make sense. Let us proceed on this assumption. Then, given 8, we have:

9*. All human agents are members of the set ‘elect-in-Christ’.

On this understanding of Barth’s doctrine of election, the relation between election and reprobation is asymmetrical. Christ takes on the sin of all humanity, becoming the Reprobate One, whose death atones for that sin, and Christ is also the Elect One whose death brings about the (derivative) election of all humanity.

The problem with this is that it seems to entail some form of universalism. But a universalism of what kind? This depends on Barth’s understanding of, among other things, free will. And it is not entirely clear whether Barth wishes to endorse a compatibilist notion of free will, or a libertarian notion of free will. Since much hangs on whether Barth holds one or other of these positions, let us define them as follows:

C1. Compatibilism:/Df.: freedom of choice means being able to actualise what one desires (more precisely, a subject S is free with respect to any action A if S desires to perform A).

L1. Libertarianism:/Df.: freedom of choice means being able to refrain from an action (more precisely, a subject S is free with respect to an action A if S could have refrained from performing A).13

This C1 definition of compatibilism means human freedom is compatible with divine determination, whereas, on the libertarian definition of L1, it is not. Let us apply these two views to Barth. Take, for example, the following passage from his Table Talk:

The decisive point is whether freedom in the Christian sense is identical with the freedom of Hercules: choice between two ways at a crossroad. This is a heathen notion of freedom. Is it freedom to decide for the devil? The only freedom that means something is the freedom to me myself as I am created by God. God did not create a neutral creature, but his creature. He placed him in a garden that he might build it up; his freedom is to do that. When man began to discern good and evil, this knowledge was the beginning of sin. Man should not have asked this question about good and evil, but should have remained in true created freedom. We are confused by the political idea of freedom. What is the light in the Statue of Liberty? Freedom to choose good and evil? What light that would be! Light is light and not darkness. If it shines darkness is done away with, not proposed for choice! Being a slave of Christ means being free.14

It is not clear, from passages like this, what position Barth endorses with respect to freedom of the will. So, in order to clarify the logic of Barth’s position, we shall set out two models of Barth’s view, the first of which is commensurate with compatibilism, and the second of which is commensurate with libertarianism. Neither of these options is consistent with Barth’s denial of universalism.

If Barth is a compatibilist then it follows that, given the foregoing:

  1. All human agents are necessarily (and derivatively) elect in Christ, the elect one, by virtue of his universally efficient atonement.

This seems the most straightforward way to understand the compatibility of 1–4 and 5–8 and 9*. Barth claims, however that, his views are not universalistic Instead, he seems to believe something like the following conjunction:

10*.  All human agents are elect in Christ, the Elect One, by virtue of his universally efficient atonement and:

  1. A human agent may reject Christ, and may, ultimately, not be saved.

This seems fallacious. For, given 1–4 and 5–8 and 9*:

  1. If a human agent is a member of the set ‘elect-in-Christ’, then a human agent will inevitably be numbered among those who are saved.

This, once again, yields a version of universalism. Perhaps another way of looking at Barth might not yield the same problems. Let us try a different tack. Barth has already allowed that:

i   Christ’s atonement is universal in scope and efficacy (from 1–4).

ii   Christ is the Elect One and therefore the sole member of the set ‘elect’, in whom all human agents are elected (from 5–8 and 9*).

iii   Christ is the Elect One whose atonement for the sin of human agents is universal in scope and efficacy, and all human agents are members of the set ‘elect-in-Christ’.

The problem is that this appears to mean that all the members of the set ‘elect-in-Christ’ will be saved, since it is not possible that the Elect One’s atonement will not be effective for all members of the set ‘elect-in-Christ’.

But why is this so? Because, as previously noted, Christ’s death is not merely potentially effective, according to Barth (as it is for, say, traditional Arminians). It is actually effective for all human agents. This understanding of Barth is reflected in the fact that he labours the point that (according to 4), the appropriation by a human agent of the benefits of Christ’s death now is a matter of coming to realise that one is already saved, and in the light of that knowledge, turning from sin to salvation. (Recall also, Barth’s claim that the attempt to reject God is pointless, since, ‘this is the very goal which the godless cannot reach, because it has already been taken away by the eternally decreed offering of the Son of God to suffer in the place of the godless, and cannot any longer be their goal.’15)

Therefore, on Barth’s model, Christ’s atonement is both universal in scope and efficacy, and potentially and actually universally effective. But if Christ’s atonement is like this, and if election is in Christ, the Elect One, then there are no reprobate humans. For on such a view, God has decreed the election and reprobation of Christ; the atonement by Christ’s death; and that that atonement be actually, universally effective for all human agents. This seems to require theological determinism as its concomitant. But Barth wants to affirm 11 which appears to reflect libertarianism.

It appears, logically speaking, that he cannot affirm both 10* and 11 without incurring a contradiction. From what we have just seen of Barth’s position, either all are (derivatively) elected in Christ, whose atonement is universally effective, and this is decreed by God, or it is only potentially universal in scope and agents are free not to opt into God’s saving work in Christ. Barth denies his own position entails the first conditional here. But even this second conditional cannot work, because Barth has already conceded that Christ’s election as the Elect One means that all human agents are elect in Christ (they are not merely ‘potentially’ elect, or some such conditional, depending upon the free choice of each individual agent). If all are elect in Christ, then no meaning can be given to the affirmation of 11.

That is, on a libertarian reading of Barth’s argument, what he says appears incoherent. So Barth is caught on the horns of a dilemma (call it, the universalism dilemma). Either his view is a species of necessary universalism (via theological determinism), the view that necessarily, all people will be brought to salvation, and is coherent, or it is a species of contingent universalism (via libertarianism), the view that as a matter of contingent fact, all people will be saved, and is incoherent. Either way, his view does not appear to be logically consistent with his denial of universalism.

However, there is one other way in which Barth could be understood on this matter. It could be that Barth believes in a universal atonement that is not efficient for ail humanity. Then he might be able to pass through the horns of the universalism dilemma like this:

10**          All human agents are elect in Christ, the Elect One, by virtue of his universally unconditional atonement, and,

11*            A human agent may deny Christ’s atonement, and may ultimately not be saved.

Barth seems to suggest at certain points in Church Dogmatics that the atonement may be universal in scope, and yet be opted out of, by human agents. For instance:

If he [the believer] believes in Him, he knows and grasps his own righteousness as one which is alien to him, as the righteousness of this other, who is justified man in his place, for him. He will miss his own righteousness, he will fall from it, it he thinks he can and should know and grasp and realise it in his own acts and achievements, or in his faith and the result of it. He will be jeopardising, indeed he will already have lost, the forgiveness of his sins, his life as a child of God, his hope of eternal life, if he ever thinks he can and should seek and find these things anywhere but at the place where as the act and work of God they are real as the forgiveness of his sins, as his divine sonship, as his hope, anywhere but in the one Jesus Christ.16

Passages such as this appear to mean that Christ’s death applies unconditionally to every human being, such that all human beings are justified (as per 4). Points 10** and 4 together mean that the atonement is still applied to a particular human agent on the condition that it has not been consciously refrained from by that particular human agent.

This, however, will not work as a solution to the universalism dilemma Barth faces, for two reasons. First, it is not clear what this view means. Barth does not mean simply that the atonement is universal in scope (potentially all human agents could be saved), but not necessarily in its effect (actually, not all human agents are saved, because they have libertarian freedom and some choose not be saved). This looks like a traditional Arminian view of the atonement. Instead, he means something more like: the atonement applies unconditionally to all humanity, such that all humanity is justified before God, but, given 4, the atonement is only applied on the condition that it has not been consciously refrained from, or opted out of, by a human agent. The problem here is that this is not a solution to the universalism dilemma that is any more coherent than the alternatives already outlined. For, if this is Barth’s position, then a person can be both derivatively elect in Christ, such that their sins are atoned for by Christ, whose death has already justified them before God (as per 4), and, at the same time, be able to opt out of this justification and election in Christ, which appears, prima facie, to be contradictory. For how can a person be both justified and (derivatively) elect, and yet be able to reject that status?

There are two strengths to Barth’s position here. He could mean, (a) a person might be justified and derivatively elect at one moment, and reject that status the next. Or he could mean, (b) a person can be in a state where they are both justified and rejected simultaneously. But how could a person be both justified and rejected at one and the same time? This seems to be confused, if not incoherent. To illustrate: it would be a strange state of affairs indeed if a subject was ennobled by their king, and given a place of prominence at court, and at the same time turned out of the court, and banished from the realm. But this is what (b) amounts to.

Nor does (a) fare any better. The problem with (a) is that it makes for a strange doctrine of election if one can opt out of election, and the status it confers on an individual, at any moment. What is more, this does not seem compatible with Barth’s position laid out in 5–8 and 9*.

Secondly, this does not seem to be in agreement with what Barth says elsewhere, about the pointlessness of persisting in rebelling against God, when one’s election is secure (as per CD II/2, 319). If a person’s election is guaranteed in such a way that any rebellion against that election is a pointless exercise, it is not clear what Barth can mean by saying, in addition to this, that election is conditional. But, from what we have seen, he does appear to make both these statements in his Church Dogmatics.

So, this attempted solution to the universalism dilemma will not work. The dilemma remains intact.

Barth’s way out?

Barth himself seems to realise the logic of his own position and retreats from its consequences. Despite his comments on the scope of election and reprobation that we noted earlier, he says:

Even though theological consistency might seem to lead our thoughts and utterances most clearly in this direction [that is, the direction of universalism], we must not arrogate to ourselves that which can be given and received only as a free gift.17

In commenting on this, Joseph Bettis maintains that this is the reason why those who understand Barth’s position to be universalistic have fundamentally misunderstood him. ‘They contend’, he says:

that while Barth has allowed for temporary estrangement, he has excluded the possibility of permanent rejection through his insistence on the completeness of the atonement of Jesus Christ. Barth’s reply is clear. The threat of eternal rejection remains a real threat.18

But far from being an adequate reply to his critics, Barth’s response appears to be simply the recognition that his own views would, if he were ‘theologically consistent’, lead in the direction of universalism. Barth, however, is happy to withhold this requirement of theological consistency, because he deems that such a move would compromise divine freedom (recall his reference to receiving grace as a ‘free gift’, above). Bettis is even willing to go as far as to say, ‘Barth does not reject universalism because the future of the pagan is uncertain. He rejects universalism because the future of all men is uncertain.’19 But if this true, then Barth’s attempted way out, via divine freedom, yields a contradiction. We can express this as follows, using the i–iii propositions stated earlier:

i   Christ’s atonement is universal in scope and efficacy (from 1–4).

ii   Christ is the Elect One and therefore the sole member of the set ‘elect’, in whom all human agents are elected (from 5–7).

iii   Christ is the Elect One whose atonement for the sin of human agents is universal in scope and efficacy, and all human agents are members of the set ‘elect-in-Christ’.

But what Barth is claiming at this juncture in his argument is something like:

iv   Because God is free, the eschatological destiny of all humanity is uncertain.

The problem is that iv simply does not appear to be consistent with i–iii. In fact, it seems to contradict i–iii. One cannot consistently hold both that all humanity have been (derivatively) elected, so that all their sin has been efficaciously atoned for by Christ, and that the soteriological status of all humanity is uncertain, any more than one can hold both that all the Conservative candidates fielded have been elected to Parliament, so that they may all return to their offices in the Palace of Westminster, and that the future candidacy of all Conservative parliamentary candidates fielded is uncertain. Either their future candidacy is uncertain, or it is not. If they have been returned to Parliament, then their future candidacy simply cannot be uncertain.20Similarly, either the question of whether all humanity are (derivatively) elect and efficaciously atoned for by Christ is uncertain, or it is not. If all humanity have been (derivatively) elected and efficaciously atoned for by Christ (as per i–iii), then their soteriological status simply cannot be uncertain (as per iv). This seems fatal to the consistency of Barth’s position.21

1 Eberhard Jüngel, Karl Barth, A Theological Legacy, (trans.) Garrett E. Paul (Philadelphia: Westminster Press 1986), 44–45. Compare also Church Dogmatics IV/III, 477–78.

2 John Colwell, ‘The Contemporaneity of the Divine Decision: Reflections on Barth’s Denial of Universalism’ in Nigel Cameron (ed.) Universalism and the Doctrine of Hell (Carlisle: Paternoster, 1992), 160.

3 Joseph D. Bettis, ‘Is Karl Barth a Universalist?’, in Scottish Journal of Theology 20 (1967): 423. Compare George Hunsinger, How to Read Karl Barth (Oxford: Oxford University Press, 1991), 132.

4 Hans Urs von Balthasar, The Theology of Karl Barth, (trans.) John Drury (New York: Holt Rinehart and Winston, 1971), 163.

5 G.C. Berkouwer, Divine Election, (trans.) Hugo Bekker (Grand Rapids: Eerdmans, 1960), 229.

6 Geoffrey W. Bromiley, The Theology of Karl Barth (Grand Rapids: Eerdmans, 1979), 97.

7 See Jonathan Kvanvig, The Problem of Hell, (New York: Oxford University Press, 1993), 74, for these distinctions.

8 Church Dogmatics II/2, 123, (hereinafter CD, followed by volume and part-volume numbers, and pagination, as is standard in making reference to Barth’s Dogmatics). In commenting on this passage in Barth, Colin Gunton observes that, ‘the cross is a substitutionary bearing by God in Christ of God’s rejection of human sin. Barth can speak of the one rejected, because through Jesus’ rejection the rejection that the human race has merited is taken away’. Colin Gunton, ‘Salvation’, in The Cambridge Companion to Karl Barth, John Webster (ed.) (Cambridge: Cambridge University Press, 2000), 145.

9 CD II/2, 164.

10 CD II/2, 271.

11 See CD II/2, 317 ff. The two citations are from George Hunsinger’s How To Read Karl Barth, 130, my emphases. Hunsinger points out the unconditional nature of the Barthian formula, observing, ‘since, in Barth’s understanding, God has already freely included us [in salvation], it falls to us henceforth freely to receive our inclusion as the gift it is proclaimed to be’. Ibid., 130–31.

12 Karl Barth CD II/2, 318–19. Of this view Bruce McCormack has recently commented, ‘[t]aken on the most superficial level, the revolution which Barth effected in the Reformed understanding of predestination was to replace Calvin’s version of double predestination with a universal election … Jesus Christ is the Subject of election and its Object, the electing God and the elected human. That is the fundamental thesis which shapes the whole of Barth’s doctrine of election.’ From ‘Grace and Being: The Role of God’s Gracious Election in Karl Barth’s Theological Ontology’ in John Webster (ed.) The Cambridge Companion to Karl Barth, 93.

13 I owe these definitions to Dr Myron Penner.

14 Karl Barth, Table Talk, (ed.) John D. Godsey, Scottish Journal of Theology Occasional Papers No. 10 (Edinburgh: Oliver and Boyd, 1963), 37.

15 CD II/2, 319, cited above.

16 CD IV/1, 631. I am grateful to Dr Myron Penner for this reference.

CD Church Dogmatics

17 CD IV/3, 477.

18 Bettis, ‘Is Karl Barth a Universalist?’, 433.

19 Ibid.

20 Of course, a parliamentary candidate who has been returned to parliament might withdraw from their position, or may withdraw at some later date. But that is beside the point. My claim here is about the consistency of saying the future candidacy of a particular Conservative MP is at one and the same time a settled question (they have been returned to Parliament), and an open question (it is not clear whether this MP has retained their seat or not). Whether that MP, having been re-elected, then withdraws or not, is a separate issue.

21 My thanks to Prof. Paul Helm, Rev. Prof. Alan Torrance, Dr Myron Penner and Mr Stuart Noble for helpful discussions on this topic.

Oliver D. Crisp

University of Notre Dame