The camera is in tight on the actress's face. The scene calls for her to move through a sequence of emotions in twenty seconds — recognition that her lover has returned, jealousy at realizing he is with someone, anger at the visible betrayal, momentary recovery to composed-disdain, and then sudden tears. Each beat lasts three or four seconds at most. The whole scene rests on what her eyes do; her body holds still, the dialogue is sparse, the cuts will be on her face.
She moves through it: Snigdhā, the loving glance — eyeballs still, slight tears already there from anticipation. Then Vismitā, the astonished — fully blown, eyeballs turned up, eyelids motionless — when she sees the other woman. Then Bhayānvitā for an instant, the awe-stricken — widely opened, eyeballs mobile and away from center — as the situation registers. Then Kruddhā, the angry — rough, eyelids motionless and drawn up, eyeballs immobile and turned up, eyebrows knitted. Then a brief Dṛptā, the haughty — steady and widely opened, eyeballs motionless, showing prowess — as she gathers composure. Then Karuṇā, the pathetic — upper eyelid descended, eyeball at rest, gaze fixed at the tip of the nose, tear.
Six glances. Twenty seconds. The audience reads each transition without conscious recognition that they have just received six distinct named operations. The actress is doing what NS VIII.38-91 names: deploying specific eye-configurations from a catalog of thirty-six.1
NS organizes the thirty-six into three groups:2
Eight rasa-glances (NS VIII.38, 44-51) — one per rasa: Kāntā (Erotic), Bhayānakā (Terrible), Hāsyā (Comic), Karuṇā (Pathetic), Adbhutā (Marvellous), Raudrī (Furious), Vīrā (Heroic), Bībhatsā (Odious). These produce the audience's experience of the rasa — the eye-configuration that makes the audience's body register the corresponding aesthetic state.
Eight dominant-state glances (NS VIII.39, 52-60) — one per sthāyibhāva: Snigdhā (love), Hṛṣṭā (joy), Dīnā (sorrow), Kruddhā (anger), Dṛptā (pride), Bhayānvitā (fear), Jugupsitā (disgust), Vismitā (astonishment). These encode the character's interior state — what the character is feeling, separate from what the audience is experiencing.
Twenty transitory-state glances (NS VIII.40-42, 61-81) — twenty specific glances for the most-stageable of the thirty-three vyabhicāri-bhāvas: Śūnyā (vacant), Malinā (pale), Śrāntā (tired), Lajjānvitā (bashful), Glānā (lazy), Śaṅkitā (apprehensive), Viṣaṇṇā (dejected), Mukulā (the half-closed-bud), Kuñcitā (contracted), Abhitaptā (distressed), Jihmā (crooked/sly), Lalitā (amorous), Vitarkitā (conjecturing), Ardhamukulā (half-bud), Vibhrāntā (confused), Viplutā (disturbed), Ākekarā (half-shut), Vikośā (full-blown joyful), Trastā (frightened), Madirā (intoxicated).3
The three sub-groups serve three different operational purposes. The rasa-glances are what the audience receives — the configuration that produces the rasa-effect. The dominant-state glances are what the character occupies — the interior state being encoded. The transitory glances are texture-of-the-moment — the emotional sub-states the character moves through within a sustained dominant state.
The framework's thirty-six is unusually fine-grained. Modern screen-acting typically operates with eight to ten generic eye-acting moves; the framework's resolution is over three times finer. Camera-proximity verifies the framework's claim: the audience's body distinguishes a Kāntā glance from a Hāsyā glance from a Karuṇā glance long before the conscious mind names what is happening. Eye-acting is a real channel operating at high resolution that contemporary craft-vocabulary mostly does not match.
The rasa-glance / dominant-state-glance distinction is theoretically significant. They are different layers — the character's interior and the audience's reception — and the actor's craft is to deploy both simultaneously. The actress in the example above is producing the Vismitā configuration (her character is astonished) while also producing the Kāntā configuration's aesthetic register (the audience is being moved toward Erotic-related rasa-content because the scene is about love-betrayed). Skilled performance manages both layers.
The transitory-state glances are the actor's texture vocabulary. Sustained sorrow looks one way; sustained sorrow with a moment of conjecture (Vitarkitā) looks different; sorrow with a moment of bashfulness (Lajjānvitā) looks different again. The twenty transitory glances give the actor the working palette for moment-by-moment texture-variation within sustained emotional states.
The thirty-six are the most directly transferable element of NS for close-up screen acting:
As close-up acting craft. Camera close-ups reveal eye-acting at its finest. Explicit thirty-six-glance vocabulary lets the screen-actor deploy specific patterns deliberately rather than producing eye-work that reads as approximate.
As animation reference. Animators benefit from explicit eye-acting libraries with body-mechanics-and-uses attached. The framework's thirty-six is one of the most developed working catalogs available.
As reading-craft for directors. Directors observing actors can use the framework to name what they are seeing and to request specific adjustments ("more Vīrā in the resistance, less Raudrī") with precision that abstract direction does not produce.
As anti-flatness diagnostic for close-up performances. Eye-acting that operates with three or four default configurations reads as flat at close-up resolution. The framework's thirty-six is the variety the audience's body can read; deploying only a handful leaves the rest of the eye-vocabulary unused. Recovery would refine close-up acting craft substantially.
Resolution claim: thirty-six is fine. Whether modern audiences distinguish all thirty-six at conscious level is unclear; whether their bodies distinguish them is more plausible and empirically testable through modern micro-expression research.
Cultural specificity: eye-acting conventions are partly culture-specific. The framework's thirty-six may need testing for cross-cultural transfer. Some of the configurations (especially the gender-coded Madirā, the intoxicated glance) carry cultural-specific connotations.
Translator-instability: as with other gesture-frameworks, glance-names translate variably across English Sanskrit poetics. The operational definitions in Ghosh's translation are clearer for some glances than others.
Bharata is the single primary source for the thirty-six, and the productive tension is between systematic-completeness (thirty-six covers the eye-acting field as Bharata's tradition observed it) and internal-asymmetry (the rasa-glances and dominant-state glances correspond to existing eight-element catalogs; the twenty transitory-glances are a selective subset of the thirty-three vyabhicāri-bhāvas rather than covering them all). The framework treats some transitory-states as eye-actable and others as not — the selection is doctrinal.
This is structurally significant. NS could have given thirty-three transitory-glances to match the vyabhicāri-bhāva catalog exactly. It gives twenty. The implicit claim is that thirteen of the transitory states do not have distinctive eye-configurations that the audience reads at the resolution required — they manifest through other body-channels (gait, voice, hand-gesture) rather than through the eyes specifically. The selection-down to twenty reflects what the dramatic tradition observed about which emotional-textures the eyes actually encode versus which emotional-textures other body-parts encode.
The framework's resolution is the working resolution of eye-acting specifically. Other body-parts have their own catalogs with their own resolutions (the sixty-seven hand-gestures, the thirteen head-gestures, the gait-vocabulary). Each catalog is sized to what that body-part actually distinguishes; the eye-vocabulary is thirty-six because the eye operates at that resolution.
The thirty-six-glances framework is a creative-practice page, but the underlying claim — that the eye-region produces specific rasa-encoded and state-encoded configurations that audiences read at involuntary-level — runs into adjacent vault domains.
behavioral-mechanics: Face-Reading — Siang Mien — Chinese face-reading and modern micro-expression research (Ekman, FACS) catalog eye-region configurations for detection-purposes. The structural parallel to the thirty-six is exact: both frameworks insist that eye-acting operates at high resolution and that the trained observer reads it precisely. NS provides the producer-side vocabulary (what the actor deploys deliberately); Siang Mien and FACS provide the reader-side vocabulary (what the observer detects). Combined, the two frameworks describe the full encode-decode operation of eye-region communication. The thirty-six-glance vocabulary suggests producer-resolution should match reader-resolution — if face-reading detects this many patterns, eye-acting should be able to produce this many patterns deliberately. The handshake's deeper claim: the eye-region operates as a high-resolution signal-channel with thirty-six-ish distinct configurations that the receiving nervous system reads at involuntary level, and skilled craft at either end of the loop (acting or reading) approaches this resolution while unskilled work operates at much coarser resolution.
psychology: State-Dependent Memory and the Somatic Unconscious — depth-psychology research treats eye-movement patterns as involuntary signals of interior state. EMDR therapy operates on eye-movement at therapeutic level; the basis is that eye-movement patterns are real indicators of state-and-memory accessibility. The structural parallel to NS's thirty-six glances is at the eye-as-state-signal level. Both frameworks insist that eyes operate as a body-state communication channel beyond conscious control. NS treats this as material for deliberate craft (the actor summons the inner state and the eye-configuration follows, or trains direct production of the eye-configuration through technique); somatic-research treats it as diagnostic (the patient's eye-configuration reveals state the patient may not articulate). The frameworks together describe the dual operation: skilled actors produce voluntarily what untrained subjects produce involuntarily; skilled observers read both. The handshake reveals that classical Indian acting-tradition and modern somatic-trauma research are working with the same eye-state mechanism from opposite operational angles. Modern micro-expression research (Ekman's FACS) catalogs eye-configurations at resolutions comparable to NS's; the convergence across cultures and millennia suggests the thirty-six is approaching empirical structure of the human eye-region's signal-capacity rather than imposing culture-specific convention.
If the framework's claim is correct — that thirty-six specific eye-configurations produce thirty-six specific rasa-and-state readings — modern screen-acting training has been operating with substantially less explicit eye-craft vocabulary than the medium requires. Camera close-ups reveal eye-acting at the highest resolution; the gap between actor's intentional vocabulary and audience's reading-capacity is structurally larger here than in any other body-channel. Recovery of the thirty-six would refine close-up acting craft measurably.
Generative Questions