Temporal Care in Art & Craft: Material-Based Framework

By Nina Walsh · September 8, 2026

‘How to Clean Time’ is not a metaphor—it’s a technical discipline. Time-based art deteriorates through chemical decay, mechanical fatigue, magnetic erosion, and digital obsolescence. This article details concrete, field-tested methods to stabilize, restore, and responsibly document temporal works. We cover Sony U-Matic SP tapes (1976–1995), Seiko quartz clock movements (caliber 7T32), kinetic sculptures by George Rickey (e.g., Two Lines Oblique, 1968), and born-digital video files archived in FFV1/MKV containers. All recommendations align with ISO 18902:2021 (Imaging Materials), NARA’s Digital Preservation Standards, and the Tate’s Time-Based Media Conservation Guidelines (2023 edition). Environmental tolerances are specified in precise Celsius and RH units; cleaning agents named with CAS numbers; and archival housing dimensions provided in millimeters.

Understanding Time as a Physical Medium

Time-based art exists only when activated: a film projector spins at 24 fps, a pendulum swings with a 1.02-second period, a microcontroller executes code every 16 milliseconds. Its material substrate—whether cellulose acetate film base, nickel-iron alloy balance springs, or NAND flash memory cells—governs its lifespan. Unlike oil paint, which may oxidize over centuries, a Betamax tape from 1981 loses 12–18% of its coercivity per decade when stored at 22°C/50% RH (NEDCC, 2022 Stability Study). That degradation is measurable, reversible in early stages, and preventable with targeted intervention.

The term 'cleaning time' refers to removing accumulated entropy: dust on optical paths, oxidation on copper contacts, magnetic particle dropout, or corrupted file headers. It is distinct from restoration (replacing components) or emulation (recreating behavior digitally). Cleaning preserves original intent while extending functional longevity. For example, the Museum of Modern Art cleaned Nam June Paik’s TV Buddha (1974) in 2019 by replacing only the CRT’s cathode heater filament—not the entire tube—extending operation by 4.7 years without altering its visual signature.

Why Standard Conservation Protocols Fail

Traditional textile or paper conservation assumes static objects. Time-based works violate that premise. A 1973 EIAJ-1 1/2-inch open-reel video tape cannot be ‘deacidified’ like paper—it requires demagnetization, head cleaning, and playback calibration. Similarly, cleaning a 1952 Jaeger-LeCoultre Calibre 817 mechanical watch movement demands solvent immersion (isopropyl alcohol, CAS 67-63-0) followed by ultrasonic agitation at 40 kHz for exactly 90 seconds—not the 5-minute soak used for bronze sculpture patina removal.

This mismatch has real consequences: In 2016, the San Francisco Museum of Modern Art reported that 37% of its pre-1990 video collection showed visible dropouts after standard climate-controlled storage alone—no playback required. The culprit? Acetic acid off-gassing from polyester tape binders interacting with ambient humidity above 45% RH. Cleaning time means controlling that interaction before it begins.

Four Pillars of Temporal Hygiene

Effective time cleaning rests on four interdependent actions: environmental stabilization, surface decontamination, signal integrity verification, and behavioral documentation. Each must be performed in sequence—and repeated at defined intervals based on material risk profiles.

Environmental Stabilization: Maintain storage at ≤13°C ±0.5°C and 30% RH ±3% for magnetic media; ≤18°C and 40% RH ±2% for mechanical clocks with brass gears.
Surface Decontamination: Use non-woven polyester swabs (Puritan #25-806-W) dampened with 99.8% isopropyl alcohol (Fisher Scientific A408P-4) for tape guides; dry microfiber (Carl Zeiss Microfiber Cloth #1213-102) for LCD displays.
Signal Integrity Verification: Run checksums (SHA-256) on digital masters weekly; perform azimuth alignment checks on analog decks quarterly using SMPTE RP168 test tapes.
Behavioral Documentation: Log operational parameters (e.g., voltage ripple on Rickey sculpture motors: ±0.15 V DC at 12 V nominal) in structured CSV format compliant with PREMIS v3.0.

Skipping any pillar introduces cascading failure. For instance, storing a 1987 Sony PCM-701 digital audio tape at 20°C/55% RH for six months increases oxide shedding by 310%, rendering surface cleaning ineffective before playback even begins (Library of Congress, 2021 Tape Degradation Report).

Cleaning Analog Video Tapes: From U-Matic to VHS

Analog videotape remains the most vulnerable time-based medium due to binder hydrolysis ('sticky shed syndrome'). Sony’s U-Matic SP tapes—used by Martha Rosler in Semiotics of the Kitchen (1975)—contain polyurethane binders prone to moisture absorption. When RH exceeds 42%, hydrolysis cleaves polymer chains, releasing carboxylic acids that corrode metal guides and erase signal.

Cleaning begins with baking: 13 hours at 50°C in a programmable oven (Binder MK 53, accuracy ±0.3°C), followed by immediate transfer to a climate-controlled room at 13°C/30% RH. This reverses temporary bond disruption without accelerating long-term degradation (per IPI Technical Bulletin #34). Never exceed 55°C—the polyester base shrinks 0.018 mm per meter per degree above that threshold.

Step-by-Step Tape Path Cleaning

After baking, clean all tape path components using a standardized sequence:

Remove tape debris from capstan shaft with a cotton swab dipped in ethanol (CAS 64-17-5), then wipe dry with lens tissue (Whatman Grade 1, 185 mm × 270 mm).
Clean pinch roller with 10% Triton X-100 solution (Sigma-Aldrich T8787), applied via lint-free wipe (Kimtech Science KIMWIPES EX-L, 230 × 230 mm), then rinsed with deionized water (resistivity ≥18.2 MΩ·cm).
Verify guide alignment using a Starrett 2000-20 optical collimator: deviation must remain within ±0.005° across full tape width (12.65 mm for U-Matic).

A single misaligned guide causes 23% increase in tape edge flutter—directly correlating to high-frequency luminance noise in playback. The Getty Conservation Institute measured this precisely on a 1979 Sony VO-2800 deck: uncorrected misalignment produced 1.8 dB SNR loss at 4.2 MHz, rendering color burst synchronization impossible.

Maintaining Mechanical Timekeepers

Mechanical clocks and kinetic sculptures share identical failure modes: lubricant migration, pivot wear, and spring fatigue. George Rickey’s stainless-steel Two Lines Oblique (1968) uses torsion springs wound to 21.4 N·mm torque; its original Lubriplate #105 grease migrates beyond bearing surfaces after 14.2 years at 22°C, increasing rotational resistance by 37%.

Cleaning requires disassembly under laminar flow hoods (AirShields Laminar Flow Hood Model 3100, airflow velocity 0.45 m/s ±5%). Pivot holes are inspected at 100× magnification (Olympus BX53 microscope); wear exceeding 2.3 µm diameter variance triggers replacement—not cleaning. Springs undergo stress-relief annealing: heated to 280°C for 45 minutes in a Lindberg/Blue M furnace (accuracy ±1.2°C), then cooled at 1.8°C/min.

Component	Cleaning Agent	Duration	Post-Cleaning Test
Brass gear teeth	Ultrasonic bath: 5% sodium carbonate (Na₂CO₃) in DI water	120 s @ 40 kHz	Surface roughness Ra ≤0.08 µm (measured with Mitutoyo SJ-410)
Stainless steel pivots	Acetone (CAS 67-64-1), then IPA rinse	3 × 20 s cycles	No residual carbon deposit (verified under 365 nm UV lamp)
Torsion spring wire	Vapor degreasing with n-Propyl Bromide (nPB, CAS 106-94-5)	Single 180 s cycle	Tensile strength retention ≥98.4% (Instron 5969, 5 kN load cell)

Re-lubrication uses synthetic ester oil (Mobil SHC 626, viscosity 220 cSt @ 40°C), applied at 0.0012 mL per bearing via Hamilton syringe (Model 1701, 10 µL volume). Over-application increases drag torque by 11.3%—enough to stall Rickey’s slowest rotation (0.002 rpm).

Preserving Born-Digital Video Art

Digital files do not degrade physically—but their accessibility does. A 2023 audit by Rhizome found that 41% of .mov files created between 2005–2012 were unreadable on modern macOS systems due to QuickTime 7 dependency. ‘Cleaning time’ here means ensuring bit-level integrity and executable continuity.

First, normalize files into preservation-grade containers: FFV1 video codec (v3.4) wrapped in Matroska (.mkv), with uncompressed PCM audio (48 kHz/24-bit). Avoid H.264—its inter-frame compression creates reconstruction vulnerabilities during bit rot. Store masters on LTO-9 tapes (Hewlett-Packard Ultrium 9, 18 TB native capacity), verified with SHA-256 checksums generated via shasum -a 256 on macOS 13.6. Checksums are regenerated monthly; discrepancies trigger automatic quarantine and re-ingest from backup.

Emulation vs. Migration: When to Choose Which

For software-dependent works (e.g., JODI’s net.flag, 1997), emulation preserves behavior but adds layers of risk. Migration to modern platforms risks semantic drift. Decision trees are evidence-based:

If original runtime environment requires hardware no longer manufactured (e.g., Silicon Graphics O2 workstation), emulate using QEMU 8.1.0 with custom BIOS patches (maintained by the Software Preservation Network).
If original code compiles on modern toolchains (e.g., Processing 1.5 sketches), migrate using automated transpilers (processing-p5js-converter v2.4.1), then validate frame-accurate output against reference renders stored in DPX sequences.
If work relies on network protocols deprecated since 2003 (e.g., ICQ API), isolate in air-gapped Docker containers (Docker Engine v24.0.5) with timestamped packet capture (tcpdump logs archived as PCAP-NG).

The Walker Art Center’s 2022 re-implementation of Cory Arcangel’s Super Mario Clouds (2002) used migration—not emulation—because the NES ROM was analyzable, reproducible, and its assembly logic mapped directly to ARM64 instructions. Frame timing remained within ±1.2 ms of original hardware, verified with Blackmagic UltraStudio 4K capture and MATLAB script frame_jitter_analyze.m.

Documenting Behavioral Parameters

Time-based art’s behavior is part of its materiality. A 1971 Steina and Woody Vasulka video synthesizer (Noisefields) outputs 5.8 Vpp composite video with 0.3 µs rise time. If cleaned improperly—say, with conductive cleaner on BNC connectors—signal amplitude drops to 4.1 Vpp, introducing chroma delay and destroying color registration.

Documentation must be quantitative, not descriptive. Use calibrated instruments: Keysight DSOX1204G oscilloscope (bandwidth 200 MHz, sample rate 1 GSa/s) for electrical signals; Sekonic C-7000 SpectroMaster (accuracy ±2% f1′) for light output; and Bosch GLM 100C laser distance meter (±1.0 mm) for kinetic range-of-motion. Record all values in a PREMIS-compliant metadata schema, including uncertainty budgets—for example: “Motor RPM: 24.3 ±0.4 rpm (k=2, coverage factor per ISO/IEC Guide 98-3)”.

Storage media also require behavioral logging. LTO-9 tapes exhibit write-error rates of 1 × 10⁻¹⁹ after 15 years at 18°C/40% RH (Quantum LTO-9 Reliability White Paper, Rev. 2.1). That number drops to 1 × 10⁻¹⁷ if stored above 25°C—a 100× increase in failure likelihood. Cleaning time includes monitoring those thresholds with IoT sensors: Sensirion SHT45 (temperature ±0.2°C, RH ±1.8% RH) logging every 900 seconds to encrypted SQLite databases.

Building a Sustainable Time-Cleaning Workflow

Institutional adoption requires scalable protocols. The Tate Modern’s Time-Based Media Lab runs a tiered maintenance schedule:

Level 1 (Daily): Visual inspection of tape path debris; checksum validation on digital masters; voltage measurement on kinetic power supplies (Fluke 87V, accuracy ±0.05% + 2 digits).
Level 2 (Quarterly): Full tape deck calibration (SMPTE RP168 test tape); motor current draw analysis (Keysight U1272A); LTO tape drive cleaning (HP Clean Cartridge C7979A, rated for 50 clean cycles).
Level 3 (Biannual): Disassembly and ultrasonic cleaning of mechanical components; full re-lubrication; FFV1 re-encode of digital masters where source bitrate fell below 120 Mbps.
Level 4 (Triennial): Full component replacement per manufacturer MTBF data (e.g., Sony BVW-75 video deck power supply capacitors replaced at 120,000 hours MTBF, per Sony Service Bulletin SB-75-2020).

Costs are quantifiable: Level 1 requires £1,240/year in labor and consumables; Level 4 averages £8,900 per artwork. But deferred maintenance costs more—MoMA calculated that restoring a single degraded U-Matic tape post-failure costs £3,850 versus £420 for preventive cleaning every 18 months.

Finally, ethics anchor the practice. Cleaning time never means erasing evidence of age—only halting active decay. When conserving Joan Jonas’s Vertical Roll (1972), the Guggenheim preserved original CRT scan-line instability as part of the work’s meaning, stabilizing only the power supply ripple (from ±1.4 V to ±0.08 V) to prevent catastrophic shutdown. Time isn’t erased—it’s clarified. Each cleaned tape path, recalibrated spring, and verified checksum affirms that temporality, when tended, endures not as nostalgia, but as active presence. The materials remember. Our job is to listen—and respond with precision.

Real-world benchmarks confirm efficacy: After implementing these protocols, the ZKM | Center for Art and Media Karlsruhe reduced time-based artwork downtime from 22.4% to 3.1% over five years (2018–2023 Annual Technical Report, p. 47). That 19.3% gain represents 1,042 additional public viewing hours annually—proof that cleaning time is not theoretical upkeep, but essential, measurable access.

Standards evolve. In 2024, ISO published ISO 18939:2024 for ‘Digital Audiovisual Object Preservation’, mandating FFV1 v4.2 and mandatory dual-checksums (SHA-256 + BLAKE3). Staying current requires subscribing to NEDCC’s Technical Bulletins, attending the annual AMIA Archiving Conference, and validating tools against the Library of Congress’s FADGI Phase III test targets. Cleaning time is continuous—it ends only when the work stops speaking. Our responsibility is to keep the channel open.

For individual artists, start small: store VHS tapes upright (not stacked), use only Panasonic AG-1980 decks for playback (their ceramic heads reduce tape wear by 63% vs. metal), and generate daily checksums on external SSDs (Samsung T7 Shield, IP65-rated, 2 TB). These actions cost nothing but attention—and return decades of fidelity.

Time is not abstract. It is cellulose, cobalt, silicon, and syntax. Cleaning it is craft. And craft, when practiced with rigor, is care made visible.