Before vinyl, before shellac, workshop culture kept its own soundtrack. In 2027 a demolition crew in Sheffield lifted the 140-year-old pine floor of a former cutlery grinder’s loft and found, packed between the joists, a 6 cm thick layer of sawdust mixed with hot resin used to deaden sound. Acoustic engineers at the University of Huddersfield proved that every hammer blow, every grinder’s whistle and every lunchtime fiddle tune had modulated the packing density as the screed cooled, locking a micro-stratified record of factory life into what is essentially a lignin-based phonograph disc the size of a living-room. Using a synchrotron phase-contrast micro-tomography stage and a visco-elastic inverse filter they replayed 37 minutes of 1889 workshop audio—complete with the 2 kHz ring of a treadle-operated grinding wheel—turning compacted wood waste into the largest mechanical audio archive ever decoded.
Sawdust-resin screed (“pitch-pine pudding”) was a cheap acoustic dampener: fresh pine shavings were blended with molten colophony at 170 °C and shovelled between floorboards. While the mix cooled through 120–80 °C it behaved like a very viscous liquid; airborne sound waves (80–100 dB) generated minute pressure gradients that preferentially compressed regions under rarefaction, reducing porosity by ~1 %. Because resin continued to polymerise for decades, the density micro-laminae were frozen in place, each 20–50 µm thick, forming a vertical analogue of a wax cylinder groove but extending 6 m × 6 m in area.
Reading starts by block-sampling a 30 × 30 × 60 cm monolith under –10 °C to prevent crumble. The block is sub-scanned at 0.7 µm voxel resolution on the I13 beamline at Diamond Light Source (53 keV, pink beam). Phase-contrast retrieval (Paganin algorithm) converts X-ray phase shift to density, yielding a 3-D map where air voxels = 0, resin = 1, cell-wall = 2. A 1-D “density trace” is extracted along the compression axis (vertical), sampled at 48 kHz—sufficient for audio bandwidth.
Clock recovery exploits the factory whistle. A 900 Hz steam pipe was blown at 07:00, 12:00 and 17:00; the resulting density spike appears as a sharp porosity low. Autocorrelation gives an inter-blow interval of 4 h 00 min ±30 s, aligning the trace to the 1889 work roster. One anomalous 09:47 whistle matches a documented safety drill on 3 May 1889, confirming temporal accuracy to ±1 min.
Error correction uses workshop redundancy. Grinders hit the same anvil forty times a minute; stacking 200 identical strikes suppresses random void noise, boosting SNR by 13 dB. Weak signals—such as the 400 Hz hum of a shaft-line belt—emerge after median stacking, revealing that the workshop ran 1.3 % above nominal speed, consistent with steam-pressure logs.
Storage density is impressive. A 6 cm screed stores ~120 MB of audio per m²—across the estimated 40,000 m² of surviving resin floors in northern England, the potential corpus is 4.8 TB of industrial soundscapes predating the earliest field recordings by a decade.
Restoration is nominally destructive—the monolith is cored and then vacuum-impregnated with Paraloid B-72 for stability—but the cavity is back-filled with modern sawdust-resin coloured to match, leaving the floor visually intact. Legal title follows UK waste law: the screed is construction debris once lifted; the audio, being intangible, is released under CC-BY for non-commercial use.
For acoustic archaeologists the lesson is clear: every resin-packed floorboard is a wax cylinder. Beneath the pitch and pine lies a micro-stratified soundtrack where every hammer blow and every fiddle note still reverberates, waiting for the right X-ray flash and the right visco-elastic filter to step out of the sawdust and back into the workshop air.
