The Half Outside the Boundary
界外的一半
Kant and SAE on Where the Line Falls
About this series
The SAE War Treatise treats only the 15DD side of war. The war philosophy of the 14DD side never appears there in full form — it enters only as the object being constrained, or as the direction of a slide.
This series lets the other side speak. It reads historical works of war philosophy and asks where they run alongside SAE, where they diverge, and where they must be criticised. It is not a study of those authors. It uses them to show what SAE is.
Four working rules govern every volume:
- Where we agree, the point is why we agree — where SAE's reason and theirs part company, and what each reason therefore does and does not reach.
- Where we disagree, we set out SAE's own position rather than correcting theirs.
- Where a text touches the ground — treating a person merely as a means, colonisation, enslavement, using formal legitimacy to cover domination — criticism is required.
- Where the question lies outside SAE's range, we say nothing.
"Outside SAE's range" does not mean "we happen to agree." It means SAE's criteria do not speak to that question. The SAE War Treatise is not a war manual. Whatever belongs to the execution layer — force levels, terrain, timing, organisation, logistics — belongs to soldiers. Saying nothing is not a judgement of value.
A note on volume order: volumes are ordered by thickness of interface, not by rank. This volume comes first because its overlap with SAE is the largest — not because it is the best.
A note on this volume specifically: this is the only volume in the series to which the question "does this author carry a shadow of 15DD?" does not apply. Kant is not a member of the 14DD summit. He is where this ground was laid.
Sources and citation
Quotations from Perpetual Peace (1795) follow Mary Campbell Smith's English translation (1903). Quotations from the Doctrine of Right (Metaphysics of Morals, 1797) are cited by Akademie pagination. Two version warnings are recorded at the end.
0 · Where we think the same, in three sentences
In the Sixth Preliminary Article of Perpetual Peace, Kant sets a criterion: no state at war may permit acts of hostility which would make mutual confidence in a subsequent peace impossible. The test is not "is it cruel." It is whether the road out is still there afterwards. SAE's constraint on means has the same shape.
In the same article he writes that in war neither party can be declared an unjust enemy — that form of words presupposes a juridical decision that does not exist between states; and that no punitive war between states is conceivable, since there is no relation of superior and inferior among them. SAE's Theorem Three agrees, and for a related reason.
In 1797, in §57 of the Doctrine of Right (AA VI:347–348), he pushes this to the level of a criterion, and the anchor turns inward: a state under attack may use any means of defence except those that would make its subjects unfit to be citizens — for then it would itself cease to be qualified, under international law, to exist as a person in the relations among states. The list of prohibitions that follows begins with using one's own subjects as spies, and continues with assassins, poisoners, and those employed merely to spread false reports.
One point must be noted here, because it affects how the rest of this volume is read: §57 and the 1795 article are substantially continuous; they are not in tension. Both are concerned with the same thing, and §57 makes the object of destruction more specific: what is destroyed first is one's own people. To wage war with spies, assassination, poison, and disinformation corrupts the moral and civic standing of those ordered to carry it out — and a state composed of the corrupted can no longer stand as a person among other states.
SAE's §12 holds that the defect of gui (deception at the lowest rung of the means-hierarchy) lies in its suppression of one's own side's subjecthood. These are not the same technical claim, but the direction is one: whether a means is disqualified depends on what it does to the party using it.
SAE has nothing different to say at these three points, so they are not developed here. This series does not lecture on its predecessors.
The three sections that follow are where the thinking differs.
1 · The same ground, the line drawn in a different place
1.1 Kant's position
§58 and §60 of the Doctrine of Right (AA VI:348–350) draw a set of limits:
Conquest cannot make the defeated state a colony; its people cannot be enslaved; political freedom is not lost by defeat. The victor has no right to partition its territory, no right to make the state disappear from the earth. He condemns the plundering of a conquered people, and his reason is clean: it was not that people who waged the war; the state under whose rule they lived waged it through them. And he holds that the people does not lose its original right to constitute a commonwealth.
Above this, however, he permits one thing. Against an "unjust enemy" — one whose publicly expressed maxims destroy the conditions of international peace — states may unite to deprive it of the power to do further harm, and may cause it to adopt a new constitution which by its nature is unfavourable to the inclination toward war (eine neue Verfassung annehmen zu lassen, die ihrer Natur nach der Neigung zum Kriege ungünstig ist).
The classification of this provision must be addressed first, because it is disputed in Kant scholarship.
One influential reading holds that §60 belongs principally to ius in bello and ius post bellum — it governs what may be done once a conflict and a situation of collective defence already obtain, and does not constitute a free-standing licence to launch preventive regime-change war against a state judged unjust.
This volume does not need to take a side in that dispute, for the following reason:
Whichever class §60 belongs to, the post-war disposition it permits is explicit in the text. And SAE's §18 firewall governs exactly that: what may be done after domination has been broken.
So the two address the same question — and if §60 belongs only to post-war right, the comparison is cleaner, not weaker: there is no mismatch of stage, only two different answers at one and the same place.
1.2 What §60 does not supply
One finding here does real work in the argument:
§60 prescribes no procedure. It does not require a plebiscite or referendum; it establishes no occupying authority or drafting body; it sets no timetable for external supervision; it does not even say the new constitution must be republican.
The only substantive qualification it gives is that property condition: the new constitution must be, by its nature, unfavourable to the inclination toward war.
And it retains a harder limit alongside: the people does not lose its original right to constitute a commonwealth.
So the line Kant draws here is: one may require that constitution to have a certain property; one may not cancel that people's standing to constitute itself.
A limit must be entered here: no conclusion may be drawn from silence.
That the text specifies no procedure admits at least four explanations: the theory's reach was not extended there; it deliberately leaves procedure open; it holds that procedure is governed elsewhere (domestic law, treaty, practice); or it genuinely holds procedure to be irrelevant to the right. The available material does not let us choose the fourth on Kant's behalf.
This volume therefore records only the textual fact: §60 specifies no procedure, and does specify a property condition on the constitution together with the people's right to constitute itself. It draws no inference about what that silence means.
1.3 SAE's position
The §18 firewall draws the line earlier: break the domination; do not carry out the reconstitution. The victor may dismantle the machinery of suppression. The victor may not decide, on the other side's behalf, what is built afterwards.
So on the one thing Kant permits, SAE does not.
1.4 There is no right and wrong here
One point must be settled first. Kant does not touch the ground here. No colony, no enslavement, no annihilation of the state, no partition, and the people retains its original right to constitute a commonwealth. He holds the anti-colonial line, and holds it earlier than most of his contemporaries.
So this is not a matter of his "not going far enough." It is two paths, on the same ground, parting at this point.
The reason for the parting can be named.
Kant's path runs through Recht. Recht concerns the conditions under which external coercion can be legitimate. In its vocabulary, a constitution "by its nature unfavourable to the inclination toward war" is a formal property identifiable from outside: does it place the power to declare war with those who bear war's costs? Since that property can be judged externally, bringing it about by coercion does not exceed what Recht is able to handle.
This also explains why §60 gives no procedure. If the constraint falls on a property of the constitution, then who drafts it, whether a vote is held, and how long supervision runs are not questions this path must answer — once the property obtains, the path's requirement is satisfied.
A note of precision: this is this volume's reconstruction of what the path implies. It is not a determination of Kant's intention (see 1.2 above on the four possible explanations).
SAE's path runs through cultivation (hanyu, 涵育) — the process by which a subject unfolds itself. And the direction of that process cannot be identified from outside — not because doing so is difficult, but because once it is identified from outside, it is no longer that subject's unfolding. So SAE's limit here is not "changing the constitution is a bad thing." It is that the act of changing it replaces the grammatical subject of cultivation.
But what SAE needs to establish here is harder to prove than it looks.
"So long as the character of the constitution is externally required, the subject has already been replaced" — that step is not available. At least three cases must be separated:
| Case | Has the subject been replaced? |
|---|---|
| An outside party drafts the constitutional text and imposes it | Yes. This needs no further argument |
| An outside party imposes only a minimal negative condition (e.g. the power to declare war may not rest with those who do not bear its costs) | Undetermined |
| The internal subject drafts, votes on, and revises the constitution itself, within that constraint, and may make many different institutional choices inside it | Substantial internal authorship may well be preserved |
The second and third cases cannot be classified as "replaced" simply by invoking the words "externally required."
So SAE may draw the line earlier, but it owes a separate argument: why does any external condition on the character of a constitution displace the subject of cultivation? That argument does not currently exist.
This volume therefore reduces the point to an outstanding item rather than employing it as an established inference.
What this volume can still say is this: Kant draws the line at the people's original right to constitute a commonwealth; SAE draws it at who may identify the direction. Neither line follows directly from "humanity is an end." Each requires a further stretch of road. Kant's stretch runs through Recht; SAE's runs through cultivation — and SAE's has not yet been walked to the end. That must be stated.
1.5 What this actually reveals
The same ground does not automatically yield the same line.
This sentence deserves weight, because it overturns something SAE previously held. An earlier working judgement in this project ran: a criterion resting on the ground of principle will automatically cover every face of relation, because the category "subject is an end" does not itself distinguish inside from outside, friend from enemy.
Kant is the strongest test of that judgement: he shares that ground, and at this point he reaches a different conclusion.
So the judgement must be restated:
Principle transfers; conclusions do not automatically cover.
The ground confers the standing to ask across faces of relation — you may take it to a face it never addressed. But between the ground and a determinate line on some particular face there is a further stretch of road: whether the criterion lands in Recht or in cultivation; what on that face can be identified from outside; and what is altered by being so identified. That stretch must be walked face by face; where it cannot be walked, coverage may not be declared.
And this answers what SAE's own line rests on: not the ground, but an additional requirement on the cultivation path — that the direction of cultivation may not be identified from outside. Kant's path carries no such requirement, so he is not bound by it.
It does not follow directly from "humanity is an end." It has to be stated separately, and SAE had not stated it separately before.
2 · Formal legitimacy fails against the strong
This is the only criticism in the volume, and it is not directed at Kant — he set the problem down himself.
2.1 The instrument
Perpetual Peace, Appendix II, the transcendental formula of public right:
All actions relating to the right of other human beings are wrong if their maxim is incompatible with publicity.
The argument is clean: a maxim which, the moment I state it, defeats its own purpose, must be kept secret to succeed; I cannot avow it publicly without at once provoking universal opposition. That resistance — knowable a priori — can only arise from the maxim's being wrong toward everyone.
This is a genuine instrument. It asks nothing about motive and nothing about consequence. It asks one thing: does this maxim still stand once it is said aloud?
2.2 The point of failure Kant marks himself
He states immediately that the converse does not hold:
We cannot conclude that all maxims which can bear publicity are therefore just. For anyone who has a decided supremacy has no need to make any secret of his maxims.
One sentence, and the boundary of the instrument is drawn.
2.3 What must be criticised is the use, not the tool
The object of criticism must be fixed precisely, because this is the easiest place to fix it wrongly.
The object of criticism is not Kant's instrument. To say that "a test built entirely on formal legitimacy will, where power is grossly unequal, systematically side with the stronger" would make the instrument itself the target, and it should not be.
Kant's negative publicity formula is a sufficient negative condition — it serves only to identify when an action is wrong toward others, and he states explicitly that the converse fails. An instrument that can only say "no," falling silent where it cannot say "yes," is not thereby defective.
The object of criticism is the act of treating it as sufficient:
If the negative publicity formula is mistaken for a sufficient or complete condition of legitimacy, then the domination exercised by the strong falls into its zone of non-judgement — it survives the test, because the test was never charged with issuing the verdict "legitimate."
The criterion has not erred. The error is in taking an instrument that can only say "no" for one that can say "yes."
And this misuse is not hypothetical. A power with decisive superiority can state its maxims of domination completely and without reservation, and they still stand once stated — because their standing does not depend on the other party's ignorance. It depends on the other party's powerlessness. So in a system that recognises only this one instrument, they pass.
The cleaner the form, the more thorough the concealment when it is mistaken for a sufficient condition.
This must be criticised, because it turns a person into someone who is merely notified — formally complete, and in substance merely a means.
2.4 What SAE says here
The criterion cannot rest on the form of the maxim alone. It must also rest on the character of the relation:
Does the other party still hold a position in this relation such that a response of theirs could change something?
This separates publicity from encounter. Publicity is one party speaking. Encounter is that, once spoken, the other's response can still enter the matter. The strong party's publicity is the former, not the latter.
That "publicity" is an announcement, not an encounter.
This also explains why SAE's yangmou — open stratagem, the second-highest rung in §12's hierarchy of means — is not the same as "saying it plainly." Yangmou requires the other party to participate as a clear-eyed subject, and participation means their response has a place. To lay one's cards on the table while the other has no way to respond is not yangmou. It is announcement in another form.
Kant gave the instrument, and marked where the instrument stops. SAE says: where it stops is exactly where the criterion needs to be.
3 · Whether questioning must be avowable: a different trade-off
3.1 Kant's arrangement
Perpetual Peace, Second Supplement — the only secret article in the work:
The maxims of philosophers concerning the conditions of the possibility of a public peace shall be consulted by states armed for war.
He then explains the arrangement: the state will silently invite such counsel, while keeping that fact secret — because seeking instruction from subjects (among whom philosophers are numbered) on the rules of its conduct toward other states "seems to detract from the dignity of its legislative authority." The philosophers, for their part, are to be allowed to discuss the universal maxims of war and peace freely and publicly; the sovereign should not let this class disappear or be silenced.
Six words: speak openly, be instructed in secret.
And he draws a boundary: it is not that the state must prefer the philosopher's principles to the jurist's opinions — only that he should be heard.
3.2 SAE's arrangement
Theorem Four requires that a party at war cannot-not be questioned, and the questioning must reach four levels: that questioning exists; that it reaches the core of decision; that the questioner is protected; that the questioning genuinely bears on the decision.
Measured against those four:
- Questioning: present.
- Reaching the core of decision: adjacency only — "shall be consulted" imposes a duty to consider, not a place inside the decision procedure.
- Protection of the questioner: partial — public discussion may not be forbidden, the class may not be silenced; but there is no protection against reprisal and no procedural guarantee.
- Bearing on the decision: not secured — not taking precedence is not the same as being without effect, but being heard is not the same as necessarily changing anything.
So this is a partial mechanism, unclosed.
3.3 This is a different trade-off, not a right and a wrong
The point is that Kant's concern is real, and he put it on paper: the state regards acknowledging its own need for instruction as detracting from its dignity.
He faced a genuine choice: have the mechanism, or have the mechanism avowed. He chose the former — keep the opening, at the cost that it cannot be seen. In his setting, insisting on open avowal might have meant the opening could not be established at all.
SAE chooses the latter, and its reason is likewise structural: an unavowed questioning cannot sustain the third and fourth levels. Whether the questioner is protected, and whether questioning genuinely bears on decision, both depend on the practice being visible — an invisible questioning cannot be verified, and therefore cannot be found out when it has been hollowed.
Two trade-offs, each with its own cost. Kant's cost is that efficacy is unsecured. SAE's cost is that this requirement may, in some settings, prevent the mechanism from being established at all.
No criticism is entered here, because the ground is not touched — Kant did not abolish questioning. He placed it where it could survive.
3.4 But one thing is worth recording
Even the man who laid this ground, when he wrote a questioning mechanism onto a state armed for war, hid it.
Not because he did not go far enough. Because this requirement, once it lands on an actual structure of power, meets a specific resistance — dignity. Kant is the one who wrote that resistance down, so it is from him that its shape can be seen.
Whether that resistance is present in all traditions, this series has not finished its sample, and makes no claim.
Conclusion
Three points: one divergence of path, one structure that must be criticised, one different trade-off.
They reveal one thing: SAE and Kant share a ground, and the ground does not settle everything. Descending from "humanity is an end," the choice between Recht and cultivation yields, at determinate points, different lines, different landing places for the criterion, and different trade-offs.
What this volume leaves SAE is not knowledge about Kant. It is a correction to SAE itself:
Principle transfers; conclusions do not automatically cover.
The §18 firewall does not follow directly from the ground. It depends on a further requirement on the cultivation path — that the direction of cultivation may not be identified from outside. That requirement had not been stated separately before. And its own argument is not yet complete (see 1.4).
The "half outside the boundary" in the title means this. It is not that Kant failed to write criteria for conduct in war — he wrote them. Nor that Recht cannot handle subjects — legal personality, civic standing, future mutual confidence: he handled all three.
It is the layer that remains unsettled even after all of those are protected: whether a subject, within a relation, still holds the position from which to question, refuse, revise, respond, and go on unfolding itself.
That layer lies outside the boundary of Recht. It is where cultivation takes over.
Appendix · Sources and verification status
| Material | Edition | Location | Status |
|---|---|---|---|
| Kant, Perpetual Peace | Mary Campbell Smith, trans. (1903) | Preliminary Articles 3, 5, 6; First Definitive Article; Second Supplement; Appendix II | primary source verified |
| Kant, Doctrine of Right | Akademie edition, German | §57 (VI:347–348), §58 (VI:348–349), §59 (VI:349), §60 (VI:349–350) | primary source verified |
Two version warnings:
First, §60 opens with a parenthetical qualification bearing on the fact that the right against an unjust enemy remains limited in the quality of means while being without ceiling in quantity and degree. Hastie's nineteenth-century English translation may invert the relation between quality and quantity at this point. Wherever §60 is at issue, this volume follows the Akademie German.
Second, the Doctrine of Right as a whole is subject to scholarly dispute over editorial error and paragraph attribution. Verification did not find §57–§60 within the disputed range, but that finding itself rests on secondary scholarship.
Two reductions recorded in this volume
First, the fact that §60 specifies no procedure is recorded here as a textual fact only. No inference is drawn about what that silence means (§1.2).
Second, the step from "any external condition on constitutional character" to "the subject of cultivation is displaced" is not argued in this volume, and is reduced to an outstanding item. SAE may draw the line earlier than Kant, but it needs an independent argument, and that argument does not currently exist. Until it is complete, this volume claims only that each of the two lines requires a further stretch of road; it does not claim that SAE's stretch has been walked.
One interpretive dispute, on which this volume takes no side
The classification of §60 — a free-standing ius ad bellum exception, or principally ius in bello and ius post bellum — is unsettled in Kant scholarship. §1.1 explains why neither reading affects this volume's argument.
One further note: this volume does not characterise the relation between the 1795 Preliminary Articles 5 and 6 and the 1797 §60 as "two incompatible answers" — a well-developed reconciling reading exists, and "the two contradict" is not an established result. §1 uses only the provisions of §58 and §60 themselves, and does not depend on whether that contradiction obtains.
One debt recorded, not discharged
Kant's exception for a divided state addresses the case in which a state has split into two parts, each declaring itself an independent state and claiming to represent the whole.
When both parts assert in the first person that they represent the whole, which bearer does §18's "do not carry out the reconstitution" protect? And how is an external actor to identify it, without the act of identification itself completing the legislation on the whole's behalf?
Kant supplies an answer to "when does assistance cease to be classified as interference." What he does not supply — and what SAE has not supplied either — is how to identify the legitimate bearer among competing claims of representation.
On this series' method
This volume, like the rest of the series, retains its own correction records in the body text: withdrawn propositions, narrowed claims, and reductions are marked where they occurred rather than silently repaired.
They are retained for a reason: a framework that requires "cannot-not be questioned" must be the first thing questioned; and a text that shows no sign of ever having erred gives its reader no way to judge how its determinations were reached.
A companion note, Known Limits, records the items this series has left unresolved.