Arjunavarmadeva was a king who described himself as "lamp to the family of Bhoja" — a descendant in the political and intellectual lineage of the great Paramāra ruler-aesthetician of Dhārā who had written Light on Passion two centuries earlier.1 He wrote one book, and it was not a treatise. It was a commentary — Rasikasaṃjīvanī (Elixir for the Rasika) — on the seventh-century erotic-lyric collection ascribed to Amaru (or, more likely, anthologized under his name).
This is unusual. By 1215 the rasa tradition was overwhelmingly a theorist's domain: people wrote treatises, not close readings. The aesthetic shape of single poems was inferred from theoretical principles. Arjunavarmadeva ran the operation in the opposite direction. He took one poet (Amaru, the master of erotic micro-form) and worked through the rasa-structure poem by poem, showing what rasa-theory looked like as a tool for reading rather than as a system to be defended.
His central claim: rasa is the organizing concern of Amaru's entire collection.1 The implication: a poet can be a single-rasa specialist, and a critic's job is to show how a particular collection sustains its rasa-commitment poem by poem. The book is, in Pollock's framing, a vindication of the close-reading mode within the Sanskrit aesthetic tradition.
Arjunavarmadeva opens with a problem. Amaru's collection begins not with an erotic poem but with two invocations — one to the mother goddess, one to Shiva as destroyer of the Three Cities. Neither is straightforwardly an erotic-rasa poem. Yet Amaru is supposed to be the erotic master. What is going on?
His answer reactivates a distinction going back through Mammaṭa and Ānandavardhana to the earliest layers of rasa-theory: rasa-poetry vs emotion-poetry vs subject-matter-poetry.1 Three things a poem can foreground:
Rasa-as-dominant: the aesthetic experience itself is the poem's center; figures of speech function as ornaments to that experience.
Emotion-as-dominant (called a "rasa-laden statement" — rasavat): the poet's affective attitude or the depicted emotional content is the center; rasa functions as an ornament to that attitude.
Subject-matter-as-dominant: the topic itself (the goddess's power, Shiva's grandeur) is the center; both rasa and figures function as ornaments to the topic.
The distinction sounds technical but does work. It lets Arjunavarmadeva argue that Amaru's first two poems are not failed-erotic poems — they are subject-matter-dominant poems where the rasa apparatus has been deliberately placed in subordinate position to celebrate the deity. The third poem ("The face of a woman making love on top — / hair mussed and swaying, earrings swinging") flips: now the rasa is dominant; the figures are ornaments to it.1 The collection moves from subject-matter-mode to rasa-mode.
In his analysis of the Shiva poem, Arjunavarmadeva works out the technical question of how multiple rasas can coexist in the same verse.1 The poem has:
The classical rule says contradictory rasas cannot coexist — that erotic and tragic, in particular, are mutually exclusive. Erotic-thwarted requires possibility of reunion; tragic-grief presumes finality of loss. Yet here both function within a single verse.
Arjunavarmadeva's resolution: when subsidiary rasas are blocked from full development and made subordinate to a third, they enhance rather than contradict the dominant rasa. The erotic-thwarted (jealous wives) enhances the tragic (their grief), which in turn enhances the heroic (Shiva's triumph). Even then, the heroic itself is subordinate to the subject-matter (the god's grandeur and the poet's wonder). The verse is a layered rasa architecture where each subordination is doing work.
His conclusion: the ability to compose in mixed rasas is the touchstone of a poet.1 A poet who can sustain only one rasa is a specialist. A poet who can layer rasas in coordinated subordination — that is the master. Amaru's invocations demonstrate mixed-rasa mastery before the collection settles into its single-rasa erotic mode. Arjunavarmadeva quotes "the commentator on Bharata" (likely Abhinava): "A single verse of Amaru is worth a hundred whole works."1
Embedded in the goddess-poem analysis is a defense of erotic poetry that depicts the wife of another man — a moral problem the tradition had wrestled with since the Nāṭyaśāstra. Arjunavarmadeva's answer: such poetry is itself a form of instruction, since "this is how a woman's character can be destroyed — so she must be carefully guarded."1 He cites Vātsyāyana's Kāmasūtra as precedent: a man becomes immune to deception about his own wife by familiarizing himself through an authoritative work with the ways of adultery.
The argument is a remarkable inversion: erotic literature is not just morally permitted but morally necessary — its instructional value lies in its accurate depiction of how seduction works. The reader's protective stance toward his own wife depends on his having read about other men's wives. Pollock notes this is the standard medieval defense of erotic literature, made unusually explicit here.
The framework's internal tension: Arjunavarmadeva uses rasa-theory to read Amaru, but the conclusions sometimes invert the theory's standard hierarchy. The classical assumption is that rasa is the highest aesthetic value — "the soul of poetry," as Ānanda put it. Yet Arjunavarmadeva repeatedly shows poems where rasa is SUBORDINATE to subject-matter, and these are not defective poems but exemplary ones. The implication is that even the "soul of poetry" can take secondary position when the work's overall design demands it. This is harder to reconcile with strict rasa-supremacy than Arjunavarmadeva acknowledges.
A second tension: his close-reading method is a discipline of attention, but his close-readings reach foreordained conclusions. The first poem MUST be subject-matter-dominant because it is an invocation. The third poem MUST be rasa-dominant because it is an erotic verse about love-making. The framework partly precedes the readings rather than emerging from them.
Arjunavarmadeva and Ānandavardhana, separated by 340 years, share the conviction that rasa is the deep grammar of poetry. They diverge sharply on what to do with that grammar critically. Ānanda built a theoretical apparatus (dhvani, propriety, manifestation) and applied it programmatically. Arjunavarmadeva builds nothing new theoretically — he uses Ānanda's apparatus as a toolkit for close reading. The convergence shows: the same theoretical inheritance can produce either system-construction (Ānanda) or system-deployment (Arjunavarmadeva). The split reveals what was missing from the theory-heavy mainstream: the disciplined act of running the theory through actual verses to see what each verse is doing. The Sanskrit tradition produced relatively few works in Arjunavarmadeva's mode. That scarcity is the tradition's loss.
Arjunavarmadeva and Bhānudatta (c. 1500), separated by 285 years, both worked in the post-treatise mode of using rasa-theory to ground specific poetic projects. Bhānudatta wrote a treatise that illustrated theory with his own poetry. Arjunavarmadeva wrote a commentary that drew theory out of someone else's poetry. The two represent the discipline's two non-treatise modes: theory-illustrated-by-original-poetry vs theory-deployed-on-existing-poetry. The convergence: both move the tradition's center of gravity from abstract theory toward concrete poetic engagement. The split: Bhānudatta produces a portable framework; Arjunavarmadeva produces a tradition of close-reading practice. The tradition would have benefited from more Arjunavarmadevas — readers who showed what the apparatus could do on actual poems.
Arjunavarmadeva's contribution is the Elixir for the Rasika as a model of close reading within the rasa tradition — a way of bringing the system's vocabulary into contact with the unrepeatable specificity of individual poems. That move has direct adjacents in two contemporary domains: literary-craft pedagogy (where the relationship between theory and close reading is a perpetual live problem) and AI-collaboration (where the question of how to evaluate generated text in light of an aesthetic framework has only just become urgent).
Creative-Practice: Non-Fiction Writing Craft Hub — Arjunavarmadeva's rasa-poetry / emotion-poetry / subject-matter-poetry distinction is structurally identical to a contemporary craft-pedagogy distinction that lacks a settled vocabulary: is THIS piece centered on the experience it produces in the reader, on the emotional charge the writer brings to the topic, or on the topic itself? Most contemporary essays drift among these three centers without the writer noticing the drift. The handshake reveals: Arjunavarmadeva's three-mode framework, applied as a pre-writing diagnostic, would force writers to commit to one center per piece and treat the other two as subordinate. The same essay reads completely differently depending on whether the writer treats it as rasa-mode (experience-centered), emotion-laden-mode (writer's stance centered), or subject-matter-mode (topic-centered). Editors who know which mode a piece is targeting can edit decisively. Editors who don't end up softening every center in turn and producing pieces with no center. The Arjunavarmadeva framework is exactly the missing pre-writing diagnostic. Crucially: rasa-mode (experience-centered) is not the only legitimate mode — the invocations in Amaru are exemplary subject-matter-mode poems. Contemporary craft pedagogy that treats reader-experience as the only acceptable target unintentionally narrows what writers can do.
AI-Collaboration: AI-Assisted Creator Hub — Arjunavarmadeva's "mixed-rasa touchstone" criterion (the master can compose in coordinated multi-rasa modes; the apprentice can only sustain one rasa) is a precise frame for evaluating AI-generated text. Most large-model outputs sustain a single tonal register. The model defaults to mode-collapse — comic stays comic; tragic stays tragic; erotic stays erotic. The handshake produces a Arjunavarmadeva-derived prompt-engineering criterion: the test of whether a generated piece is master-quality vs apprentice-quality is whether subsidiary tonal modes are deliberately subordinated to enhance the dominant mode rather than randomly co-occurring or being suppressed. The contemporary AI-evaluation tradition lacks this criterion. With it, we have a precise way to ask: does this generated text have RASA-LAYERING, or just a single sustained tone? The former is a sign of compositional intelligence; the latter is a sign of statistical averaging. The criterion can be operationalized in prompting (ask for explicit dominant and subordinate rasas; check whether the subordinates are functioning to enhance the dominant or to dilute it). Arjunavarmadeva's framework lets us articulate what "good writing" generated by AI is structurally doing that bad-but-fluent generated text fails to do.
Behavioral-Mechanics: The Cooking Metaphor — Spectator as Connoisseur — Arjunavarmadeva's commentary mode treats the reader as a connoisseur capable of palate-discrimination — the rasika of his title. The framework presumes the reader can taste the difference between dominant-rasa and subordinate-rasa, between rasa-mode and emotion-mode and subject-mode. The handshake reveals: the rasika is not a natural-born type. The rasika is trained by close-reading commentaries like Arjunavarmadeva's. Influence operations targeting any sophisticated audience face an analogous training problem — the audience must be cultivated into the discriminations the operation requires. Arjunavarmadeva's commentary IS palette-training infrastructure. Modern marketing and influence often assume audience-discrimination as a given; Arjunavarmadeva's framework reveals that discrimination is itself the product of cultivation, and that the most powerful influence campaigns are simultaneously palette-cultivation campaigns.
The Sharpest Implication
Arjunavarmadeva's framework destabilizes the rasa-tradition's assumption that the experience produced in the audience is always the primary aesthetic event. Some great poems are great BECAUSE they subordinate rasa to subject-matter. The invocation to the goddess is exemplary not because it produces heroic rasa but because it celebrates the goddess's power while using the heroic rasa as ornamental support. The implication: a craft pedagogy that takes "what does the reader feel?" as the only question to ask has missed roughly two-thirds of what poetry can do. The rasika is trained to ask three questions about every poem: is it about an experience, about a stance, or about a thing? — and to read accordingly. The discomfort: contemporary content design overwhelmingly defaults to the experience-question, leaving the stance-question and thing-question undertheorized. Whole traditions of writing (the invocation, the encomium, the philosophical essay) require the other two questions to be primary. Arjunavarmadeva's distinction names exactly what is being lost when we collapse all three into one.
Generative Questions