diff --git a/web/sections/SubjectFields.tsx b/web/sections/SubjectFields.tsx
index 00e29c9..1b68809 100644
--- a/web/sections/SubjectFields.tsx
+++ b/web/sections/SubjectFields.tsx
@@ -2,6 +2,7 @@ import { useEffect, useState } from 'react';
 import { useClient } from 'hooks/useClient';
 import { Input } from '@/components/ui/input';
 import { Textarea } from '@/components/ui/textarea';
+import { RichText } from '../RichText';
 import { useAutosave } from '../useAutosave';
 import type { SubjectRecord, SubjectSection } from './subjects';
 
@@ -67,12 +68,22 @@ export const SubjectFields = <T extends SubjectRecord>({
         </p>
       </div>
 
+      {/*
+        Rich text, not a textarea.
+
+        It is the paragraph the template draws under the name — the one piece of writing about this
+        thing — so it wants the same headings, emphasis, links and alignment the page's own text
+        blocks have. `Prose` on the site already renders HTML when it finds any and falls back to
+        paragraphs when it does not, so what is already stored as plain text goes on working.
+
+        Sanitised on the way in by the API, against the same allowlist the page components use.
+      */}
       <div className="space-y-1.5">
         <label className="text-xs text-muted-foreground">Description</label>
-        <Textarea
-          rows={4}
+        <RichText
           value={value('description')}
-          onChange={(event) => set('description')(event.target.value)}
+          onChange={(html) => set('description')(html)}
+          placeholder="What this is."
         />
       </div>