Musings on Saint Anselm’s Dilemma

Reflexiones sobre el dilema de San Anselmo

J.-Martín Castro-Manzano
josemartin.castro@upaep.mx

Abstract: In this contribution we suggest Resumen: En este trabajo proponemos dos

two simple contentions: i) that Saint An- afirmaciones: i) que el argumento ontoló-
selm’s ontological argument, as presented gico de San Anselmo, como se presenta en
in the Proslogion, only makes sense if em- el Proslogion, sólo tiene sentido al interior
bedded in an ordered and bounded ontol- de una ontología ordenada y acotada; y
ogy; and ii) that an interpretation of the ii) que una interpretación del argumento,
main premise of the argument, within dentro de tal ontología, produce un dilema
such ontology, produces a dilemma that que demanda una nueva revisión del argu-
demands a new revision of the argument. mento.
Keywords: Ordered ontology; bounded on- Palabras clave: Ontología ordenada; ontolo-
tology; greatness. gía acotada; grandeza.

1. Introduction

We share the opinion that the Proslogion should prove as much a philo-
sophical as it does a mathematical work of art1, for it presupposes an ontol-
ogy that requires assumptions close to order theory, namely, assumptions
of order and bounds. Following these assumptions, in this short contribu-
tion we suggest two simple contentions: i) that Saint Anselm’s ontological
argument, as presented in the Proslogion, only makes sense if embedded in
an ordered and bounded ontology (§2); and ii) that an interpretation of the

1 
Authoritative scholars do not necessarily share this opinion. Anselm Stolz (1967),
for instance, suggested the Proslogion is a piece of mystical theology. Karl Barth (1960)
offered a more conservative assessment in that St. Anselm’s work is fundamentally
theological. Éttiene Gilson (1934), however, argued the Proslogion is not philosophy
any more than it is theology or mysticism, and thus suggested the label gnosticisme
chrétien. We believe, nevertheless, that the Proslogion has some resemblance with con-
temporary classical mathematics for its central argument behaves like a theorem of
uniqueness and existence as they usually appear in classic mathematics (cf. Oppen-
heimer and Zalta, 1991).

Studium. Filosofía y Teología. Vol. XXII, 43 (2019) 05-12 5

ISSN 0329-8930 - ISSN 2591-426X
 Musings on Saint Anselm’s Dilemma

argument, within such ontology, produces a dilemma that demands a new

revision of the argument (§3).

2. Remarks on the Anselmian ontology

For the purposes of this contribution we define an ontology by a pair

〈D, R〉 where D is a non-empty domain and R is a relation defined on D. In
order to describe the Anselmian ontology, OA for short, we consider Pros.
II, and Pros. III−which contain subtle versions of the ontological argument
(cf. Malcolm, 1961) −, along with Pros. V, Pros. XV, and Mon. I, and Mon. III.
From these passages we can infer, given the uses of maius/melius (cf.
Brecher, 1974), that the Anselmian ontology is defined by the pair OA=〈B,
≥〉 where B is a non-empty bounded set of beings and the maius/melius re-
lation, namely ≥, is a partial order relation on B. We can summarize these
features with the following facts:

Fact 1. OA is a partially ordered ontology.

Fact 2. OA is a bounded ontology.

These facts provide evidence to the claim that St. Anselm’s ontological
argument only makes sense if embedded in an ordered and bounded onto-
logy2. Indeed, we could not account for the maius/melius relation without an
order relation (Fact 1); and we could not account for the notion of “that than
which nothing greater can be conceived” (or “the greatest being” for short)
without (upper) bounds (Fact 2). However, the very notion of “the greatest
being” is not naive and requires further attention because, for it to make
sense, OA must comply with two further remarks:

Remark 1. OA has at least one maximal being.

Remark 2. OA has at most one greatest being.

2 
These facts about the Anselmian ontology can also be inferred from a “hierarchy of
being” that can be found in the Neoplatonic assumption that, despite all beings share
existence, higher beings have a greater proportion of being. This idea allows us to un-
derstand what thinkers like St. Augustine, St. Anselm himself, Boethius, and Dante
were doing (McMahon, 2006). McMahon is of course correct in that, although the first
three did not have the analogia entis as an explicit intellectual instrument, the hierarchy
of being was a cultural commonplace, especially for St. Anselm, who was a thinker
with a Platonic/Augustinian imprint.

6 Studium. Filosofía y Teología. Vol. XXII, 43 (2019) 05-12

 J.-Martín Castro-Manzano

Consider, for Remark 1, that a maximal being in a bounded and partially

ordered ontology, such as the Anselmian, would be a being bx in the domain
for which there is no being by such that by≥bx. So, take an arbitrary being bi
in OA. Suppose, for reductio, that bi is not maximal. Then we can find another
being, say bi+1 (i.e. a next being), s.t. bi+1>bi (i.e. a next being that is greater
than the previous one). Now, if bi+1 is not maximal itself, we will find yet
another being bi+2 s.t. bi+2>bi+1>bi. If we continue looking for greater beings in
this way, we will have to stop eventually at a bound, call it g, corresponding
to a being bg, since OA is bounded above, by Fact 2. Therefore, it is impossi-
ble to have bi>bg for any i in OA. Hence, there is at least one being bg that is
maximal in OA.
Additionally, for Remark 2, since OA is a partially ordered ontology, by
Fact 1, we also have at most one greatest being. A greatest being in a boun-
ded and partially ordered ontology would be a being bx in the domain such
that for all by, bx≥by. So Remark 2 is clear, for suppose there are two greatest
beings bg and bk in OA, then bg≥bk and bk≥bg. But then, by the antisymmetry of
the order relation, bg=bk, that is to say, there is at most one greatest being in
OA, namely, bg.
Together, these Facts and Remarks make sense of the following propo-
sition:

Proposition 1. If bg is both the maximal and greatest being in OA, then bg is

the supremum in OA.

Proposition 1 clarifies the meaning of the expression “the greatest

being” and specifies that the domain of beings B should behave like the
closed-to-the-right interval [b0,bg]3, for any beings i in B, where bg denotes
the supremum; otherwise OA would comply with Fact 1 (order) but not with
Fact 2 (boundedness), which would imply the failure of Remarks 1 and 2.
At this point, some numerical examples will help clarify this last propo-
sition. Consider the open interval (−10,10). This interval is bounded above,
for instance, by the numbers 11, 12, …, 100. This is because upper bounds
need not belong to the given set. But in the interval (−10,10], for example,
10 is the least upper bound and, since it belongs to the set, it is also the su-
premum. Thus, the interval (−10,10) has a least upper bound (number 10)

3 
This interval could also be represented, without loss of generality, by the interval (b0,
bg], for the important bound is the one on the right.

Studium. Filosofía y Teología. Vol. XXII, 43 (2019) 05-12 7

 Musings on Saint Anselm’s Dilemma

but no supremum; while (−10,10] has both a least upper bound and a supre-
mum, namely, number 10. The interval (−10,+∞), on the other hand, is not
bounded above and it has no supremum.
To further illustrate Proposition 1 let us consider a more concrete exam-
ple. Imagine we have a domain of books, say a library, of different height,
say b1, b2 and b3, ordered with respect to their height in such a way that
b3≥b2≥b1. Clearly, we can identify “the highest book” in the library, that is to
say, that book “than which no higher book can be conceived” in that library,
namely, b3. Now, what must be stressed in this situation is that the only way
in which it makes sense to talk about “the highest book” is if the domain is
both ordered and bounded (i.e. if Facts 1 and 2 obtain). But notice that if the
domain of books is bounded, then b3 is bounded itself, since the bound of the
domain is determined by b3’s actual height, for otherwise the domain would
be unbounded. So, for instance, the intervals (−10,10] and (−10,10) are clearly
bounded above because 10 is bounded itself; had the number 10 no bounds
of its own, both intervals would behave like the unbounded interval (−10,
+∞), which would require the number 10 to behave like +∞, which is absurd,
for +∞ is not even a number. Likewise, in the case of the books, had the book
b3 no bounds of its own, the domain of books would look like the unbounded
interval (b1,+∞), which would require the book b3 to behave like +∞, which
is also absurd, for +∞ denotes the interval’s behavior, not an actual book or
entity. Hence, the book b3 is a bound of the domain because it has a bound
of its own, say its actual height denoted by the definite number 3. This is an
important step for what comes next because, suppose, for reductio, that bg is
the supremum of OA but is not bounded itself. If bg is the supremum of OA,
the domain B must behave like a closed-to-the-right interval, say (b0,bg]. But
if bg is not bounded itself, bg would behave like +∞, in which case B would
behave like the open interval (b0,+∞), but that would contradict Fact 2. We
call this the bounded-bound condition. The bounded-bound condition can be
best “viewed” with the aid of the following diagrams:

8 Studium. Filosofía y Teología. Vol. XXII, 43 (2019) 05-12

 J.-Martín Castro-Manzano

From left to right, the first diagram shows a representation of a boun-

ded space as a ball of finite radius so that for all bx, bg≥bx, i.e., the distance
d(bx,by)≤bg for any beings by and bx. This represents a situation that consi-
ders that we may admit a possible infinite set of beings in B, but B is still
bounded by bg. The second diagram represents OA as a bounded chain. In
both diagrams the domain is bounded because it is true that for all bx, bg≥bx.

3. A loop and a dilemma

At this point, nothing seems to be too controversial, especially regar-

ding the previous examples because, for one, the notion of “the highest
book” is scarcely problematic. However, what happens if we think, not
of “the highest book”, but of “the greatest being”? Let us entertain this
question for a moment. Suppose bg is the greatest being, according to OA.
Then, bg is precisely the supremum, by Proposition 1. Now, if bg is the su-
premum, bg is bounded itself, by the bounded-bound condition. However,
if bg is bounded in itself, it has some sort of limit, namely the limit defined
by its own bound (as in the example of the book, it’s very height was its
own bound); but if bg, “that than which nothing greater can be conceived”,
has bounds of any sort, could not we think of something greater than bg,
namely, something not constrained by the bound defined by the limit g?
After all, why would the greatest being have bounds of any sort?
In other words, if we say bg is the supremum, then we would have
to admit that bg has some sort of own bound or limit. If this is the case,
however, we could think of a being beyond that bound, in which case we
could extend B indefinitely (just as we can think of a number greater than
10 when considering the interval (−10,10]). On the other hand, if we say
there is some being greater than any other being, we would have to admit
that there is no supremum whatsoever. If this is the case, nevertheless,
we could extend B indefinitely and then B would be unbounded; but this
would be odd, because then there would be no supremum at all; but if the-
re is a supremum, then such being would have to be bounded, but then we
could think of a being beyond that bound again, namely, a greater being,
and then we could extend B again, and so on and so forth. However, this
turns out to be kind of paradoxical because it leads to evident inconsisten-
cies regarding the notion of “the greatest being”. To visually explain this
situation, consider the next flow chart:

Studium. Filosofía y Teología. Vol. XXII, 43 (2019) 05-12 9

 Musings on Saint Anselm’s Dilemma

The flow chart above starts by asking whether greatness has bounds
of any sort. If greatness has no bounds, it does not make sense to talk about
“the greatest being”. But, on the other hand, if greatness has bounds, the no-
tion of “the greatest being” makes sense, but then we can ask if we can think
of something above those bounds; and in case we can, then it does not make
sense to talk about “the greatest being” again; and in case we cannot, we get
into a loop and we return to the original question that starts the flow chart.
With this loop schema in mind, we can now expound what we call
“Saint Anselm’s dilemma”:

1. To avoid the loop, either we maintain the assumption that “bg is the
greatest” while dropping the notions of order and bounds, namely
OA; or we maintain such notions, namely OA, without dropping the as-
sumption that “bg is the greatest”.
2. The problem with the first horn of the dilemma, of course, is that we can-
not abandon OA without dropping the thesis that “bg is the greatest”, since
such abandonment would render the very proposition “bg is the greatest”
meaningless (due to Proposition 1), and then the Anselmian ontological
argument would be unsound (because of an ill-defined premise).

10 Studium. Filosofía y Teología. Vol. XXII, 43 (2019) 05-12

 J.-Martín Castro-Manzano

3. But then we should consider the other horn of the dilemma and drop
the claim that “bg is the greatest” while keeping OA, but this strikes us as
odd because the thesis that “bg is the greatest” seems to be both the main
premise of the Anselmian ontological argument and a natural conse-
quence of OA (due to Remarks 1, 2, and Facts 1, 2), but then the Ansel-
mian argument would be unsound (because it would be self-defeating).

Since, in both cases, the Anselmian ontological argument turns out

to be unsound, but we know the argument is logically valid (cf. Oppen-
heimer and Zalta, 1991), the problem must come from the layman use of
the notion of “the greatest being”, and so this notion has to be recast with
more attention in order to avoid any inconsistencies or loops. For instance,
an obvious objection that undercuts the previous remarks is the use of the
“bounded-bound condition”, since it seems to be an artificial and contro-
versial add-on to Saint Anselm’s original proposal. Granted. This so-called
bounded-bound condition is not to be found within the Anselmian corpus.
However, since the notion of “the greatest being” produces these artificial,
yet troublesome considerations, the bounded-bound condition must not be
discarded altogether as an alien one. It may certainly be artificial, but it is
also not strange to OA, due to Remarks 1, 2, and Facts 1, 2.

4. Concluding remarks

In this contribution we have suggested two simple contentions. The

first one, rather typical, states that Saint Anselm’s ontological argument, as
presented in the Proslogion, only makes sense if embedded in an ordered
and bounded ontology; the second one, which we think is rather controver-
sial, focuses on an interpretation of the argument within such ontology that
produces a dilemma that demands a new revision of the argument due to
the ambiguity of the notion of “the greatest being”.
The dilemma shows, we think, that unrevised assumptions of order
and bounds can render some propositions unsound if we do not handle
them with care. And this is an important and delicate issue, especially if we
consider, for instance, any conception of God as a being defined in terms of
greatness4, for such greatness would have to be embedded in an ordered

Whether it is done for worship-justification reasons (cf. Niemeyer Findlay, 1948) or

4 

proof-justification purposes (cf. Plantinga, 1974).

Studium. Filosofía y Teología. Vol. XXII, 43 (2019) 05-12 11

 Musings on Saint Anselm’s Dilemma

and bounded ontology in order to make any sense; but then we would need
to revise our order assumptions more carefully, because the notion of great-
ness, if not managed carefully and precisely, might lead to strange loops
where none should be.

References

Barth, K. (1960) Anselm: fides quaerens intellectum, trans. by Ian W. Robertson.

Richmond: John Knox Press.
Brecher, R. (1974) ‘Greatness’ in Anselm’s Ontological Argument, The Philo-
sophical Quarterly 24, 95, pp. 97-105.
Gilson, É. (1934) Sens et nature de l’argument de saint Anselme, Archives
d’Histoire Doctrinale et Littéraire du Moyen-âge 9, pp. 5-51.
Harsthorne, Ch. (1961) The Logic of the Ontological Argument, The Journal
of Philosophy 58, 17, pp. 471-473.
Malcolm, N. (1960) Anselm’s Ontological Arguments, Philosophical Review
LXIX, pp. 41-62.
McMahon, R. (2006) Understanding the Medieval Meditative Ascent. Washing-
ton: The Catholic University Press.
Niemeyer Findlay, J. (1948) Can God’s Existence Be Disproved? Mind 57.
Oppenheimer, P. and Zalta, E. (1991) On the Logic of the Ontological Ar-
gument, ed. J. Tomberlin, Philosophical Perspectives 5: The Philosophy of
Religion. Atascadero: Ridgeview.
Plantinga, A. (1974) The Nature of Necessity. Oxford: Oxford University Press.
Stolz, A. (1967) Anselm’s Theology in the Proslogion, ed. John Hick and Ar-
thur C. McGill, The Many-Faced Argument. New York: McMillan.

Recepción: 05.01.2019
Aceptación: 25.02.2019

12 Studium. Filosofía y Teología. Vol. XXII, 43 (2019) 05-12