A suspended micro-bridge reads a total thermal conductance: sample plus the parallel background of the device beams, residual gas and radiation, in series with two contacts. Recovering the sample means subtracting a separately measured background and budgeting the contacts — and the arithmetic punishes small samples. Slide the sample’s share down and watch the recovered value’s error balloon, exactly as in every differential design.
Model: background-subtraction error from the equal-variance independent-error law σ/share·√(1+(1−share)²) at σ = 1% per measurement, combined in quadrature with an ILLUSTRATIVE contact-systematic floor proportional to the dialed contact share. Real budgets measure the device background directly and bound contacts from multi-length or four-probe designs. Schematic teaching tool, not an instrument.