A poet at the Pallava court in late seventh-century South India wrote a book called Kāvyādarśa — Looking Glass of Poetry. He tells us in another of his works that his great-great-grandfather had come south from today's Karnataka. He calls himself Daṇḍin. He composed a celebrated prose poem (Avantisundarīkathā — Story of the Beautiful Woman of Avantī) whose adventure-portion (Daśakumāracarita — Adventures of the Ten Princes) circulated independently. He wrote a now-lost double-narrative poem in extended double entendre.1
And then there is Looking Glass of Poetry, which Pollock characterizes with unusual force: "one of the most influential works in the global history of poetics, probably second only to Aristotle's treatise in breadth of impact." Translated into multiple South Asian languages. Exerting influence on Recent Style Chinese poetry of the late Tang dynasty. Shaping seventeenth-century Tibetan poetics. The geographic reach is rivaled by no other Sanskrit literary-theoretical work.1
Daṇḍin uses "rasa" in three largely unrelated senses, only one of which would survive the subsequent tradition:
Sense 1: General "tasteful" or "sweet" poetic style. "A poem is defined as sweet when it has rasa. Rasa is found in both the language and the subject matter, and insightful people become intoxicated by it like bees by honey."1 This is rasa as a general property of literary quality — the texture that distinguishes good poetry from documentary prose.
Sense 2: "Sophisticated" turns of phrase. Wordings that are unusually well-turned. "Sound rasa" (phonemic similarity across words) and "content rasa" (sophistication in expression of meaning). Both are linguistic refinements that distinguish elevated from vulgar usage.
Sense 3: Technical figure of speech. The same three figures Bhāmaha had named — rasavat (rasa-laden statement), preyaḥ (affectionate utterance), ūrjasvi (haughty declaration). For these, Daṇḍin offers extensive illustration and more developed treatment than Bhāmaha provided.
The first two senses would be discarded by the later tradition. Only the third — rasa as a figure of speech with eight rasa-types as content — would survive. Daṇḍin's wider semantic range for "rasa" is the historical anomaly. By 875 CE Ānandavardhana would establish rasa as the SOUL of literature, narrowing the term to its technical meaning. Daṇḍin's loose general-quality and stylistic uses would be quietly abandoned.
Daṇḍin's treatment of rasavat establishes one of the foundational positions in rasa-theory: that rasa is the intensified emotion of the character. His illustrations are uniformly the character's direct speech, never external description. The poet beautifully describes Bhīma's anger as it becomes the violent rasa; Pururavas's craving as it becomes the erotic-thwarted rasa; the king's determination as it becomes the heroic rasa; the queen's grief as it becomes the tragic rasa. In each case, the rasa is unambiguously in the character.
Abhinavagupta three centuries later would identify Daṇḍin's position with "the most ancient authorities" — and indeed it is the position Bhaṭṭa Lollaṭa would systematically defend in the early ninth century. The character-locus framework that Bhaṭṭa Nāyaka would eventually overturn has Daṇḍin as its first articulate exponent.
Daṇḍin's commentator Ratnaśrījñāna (c. 950, a Buddhist monk from Sri Lanka who also translated the Looking Glass into Sinhala) is even clearer: an emotion's "achieving the highest stage" of intensification means it has been made manifest through verbal reactions, since emotions are internal states otherwise unknowable. The rasa is the manifest, intensified form of the character's stable emotion. The reader is not yet in the picture.
The Looking Glass attracted two important commentators whose work survives:
Ratnaśrījñāna (c. 950) — Ratnaśrī (Ratna's Glory) — a Buddhist monk from Sri Lanka. He translated the entire Looking Glass into Sinhala, making Daṇḍin's framework available to one of the major South Asian literary traditions that would not otherwise have had access. The Sinhala translation became the foundation for Sri Lankan classical poetics.
Vādijaṅghala ("Gazelle Among Disputants") — Śrutānupālini (Guarding the Tradition) — c. 950. A scholar at the Rāṣṭrakūṭa court of Krishna III. His actual name is unknown; the sobriquet ("gazelle among disputants") is what survived.1
Pollock praises Ratnaśrījñāna's commentary as offering "a clear and simple explanation" of how an emotion becomes rasa — explicit articulation that emotions are internal states made manifest by reactions, the kind of theoretical clarity that the post-revolutionary tradition would obscure rather than clarify.
Daṇḍin's threefold use of "rasa" represents a moment of conceptual instability that the later tradition resolved by narrowing the term. The general-style and stylistic-refinement senses — "this poem has rasa" meaning "this poem is well-written" — get absorbed by the more general category of guṇa (language qualities). The figure-of-speech sense — "this is a rasa-laden statement" — gets absorbed and transformed by Udbhaṭa into the full rasa-typology, then re-organized by Ānandavardhana as one element among others within the dhvani framework.
The transformation is one-directional. Daṇḍin's broad use of "rasa" as a general quality-term is never revived. By the time Mammaṭa codifies rasa-theory in 1050 CE, the threefold sense has been compressed into the single technical sense, and Daṇḍin's looser usage looks like a category error. But within his own framework it was not a category error — it was a careful distinction between three different operations of language that the later tradition collapsed.
A second tension: Daṇḍin's Looking Glass was translated into Sinhala, Tibetan, and Chinese — but his expansion of the rasa concept beyond figure-of-speech was not carried with the translation. The exported framework was largely the figure-of-speech catalog (which translates straightforwardly into other traditions' rhetorical analyses). The expanded "rasa = general sweetness" sense did not export.
Daṇḍin and Bhāmaha (predecessor, c. 650) converge on treating rasa as a figure of speech with the same three subtypes (rasa-laden, affectionate, haughty). They diverge on the broader semantic range Daṇḍin permits — Bhāmaha keeps "rasa" tightly tied to the figure-of-speech sense; Daṇḍin allows the broader stylistic-quality and general-sweetness usages. The tension reveals what each scholar was doing: Bhāmaha was trying to specify the discipline of poetics narrowly; Daṇḍin was trying to capture the wider semantic life of the term in actual literary practice. The post-Daṇḍin tradition agreed with Bhāmaha's tighter usage — but Daṇḍin's wider usage may have been closer to how working poets actually talked about rasa.
Daṇḍin and Bhaṭṭa Lollaṭa (c. 825), separated by 125 years, share the firm conviction that rasa is in the character — Abhinavagupta would explicitly cite Daṇḍin's framework as the precedent for Lollaṭa's utpatti-vāda (production view). They converge on the character-locus framework that the post-Bhaṭṭa-Nāyaka tradition would overturn. The convergence is structural — both treat rasa as an emergent property of the character's intensified stable emotion. The split reveals where the limit of pre-revolutionary rasa-theory lay: a clear theory of how rasa exists in the character; no clear theory of how the reader experiences it. Bhaṭṭa Nāyaka's question — what makes Rama's emotion available to the reader, who is not Rama? — was simply not asked in this framework.
Daṇḍin's framework operates at the historical baseline of post-Bhāmaha poetics and represents the position the entire subsequent tradition built upon and against.
Creative-Practice: Non-Fiction Writing Craft Hub — Daṇḍin's threefold use of "rasa" — general sweetness, stylistic refinement, technical figure — captures three different things contemporary craft instruction also wants to mean by "good writing": that it has a certain quality of sweetness (whatever distinguishes elevated from flat prose), that it shows technical sophistication (well-turned phrases, vivid metaphors), and that it produces specific emotional effects (the precise calibration of mood that makes a scene "work"). The handshake produces a recognition: contemporary writing instruction implicitly uses a Daṇḍin-style threefold rasa-vocabulary without the conceptual infrastructure to distinguish the senses. When a writing instructor says "this prose has rasa," they might mean any of the three — and the resulting feedback is correspondingly unclear. Daṇḍin's distinction, if revived, would clarify what kind of revision a piece actually needs.
AI-Collaboration: AI-Assisted Creator Hub — Daṇḍin's "sound rasa" — phonemic echoes across words producing a particular kind of stylistic pleasure — is the easiest aspect of poetry to generate computationally (rhyme, alliteration, meter are pattern-completable). His "content rasa" — sophistication of meaning — is harder. His figure-of-speech rasa — the precise calibration of emotion through aesthetic elements — is hardest. The three-tier framework provides a diagnostic for what current AI systems do well, what they do moderately, and what they fail at. Sound rasa: trivial. Content rasa: improving. Figure-of-speech rasa with proper rasa-architecture: unreliable. The handshake reveals that current AI prose generation is operating closer to Daṇḍin's first sense than to his third sense — producing surface stylistic effects without the precise emotional engineering that the later rasa-tradition would require.
Eastern-Spirituality: Ānandavardhana and the Dhvani Revolution — Daṇḍin's threefold rasa-vocabulary is what Ānandavardhana would compress into the single technical sense by transforming rasa into the SOUL of poetry. The transformation required eliminating Daṇḍin's broader senses — rasa-as-general-sweetness and rasa-as-stylistic-refinement — so that rasa-as-the-soul could mean ONE thing. Pollock argues this compression was costly: it absorbed the wider semantic life of the term and made rasa-theory artificially narrow. The handshake reveals the cost of conceptual consolidation: by making rasa technical, the tradition gained precision but lost the broader range of phenomena Daṇḍin's wider usage captured.
The Sharpest Implication
Daṇḍin's Looking Glass of Poetry shaped literary traditions across three continents — South Asia, East Asia (Tang dynasty Recent Style Chinese), and the Tibetan world — for centuries. It is the second-most-influential poetics work in world history after Aristotle. And yet most contemporary education in literature does not even encounter Daṇḍin's name. The intellectual influence of Sanskrit poetics outside the Sanskrit-reading world is one of the great forgotten facts of literary history. If Daṇḍin's framework genuinely shaped how Tang-dynasty Chinese poets understood their craft, then the standard "Western tradition" narrative of poetics is missing a major node — and the questions Daṇḍin raised (the threefold rasa, the figure-of-speech treatment) may be available as live alternatives to the framework most contemporary education in poetry actually uses.
Generative Questions