Presets
Playback
Unknown load
Mass on the short arm and the length of that arm.
Sliding poise
Slide the poise along the graduated long arm.
Beam and knife edge
Centre-of-mass offset, pivot height, damping and friction.
Layers
Model checks
The science
The instrument
A steelyard (Roman balance) is a lever with unequal arms. The unknown mass M hangs from a hook a short distance a from the knife edge; a fixed poise of mass m slides along the long, graduated arm to position x. The beam is level when the torques match: m x + m_b c = M a, so M = (m x + m_b c) / a and the scale reads mass directly. The mechanical advantage is x / a.
Why the beam tilts and how fast
The knife edge sits a height h above the beam so the beam is stable. With the beam tilted by theta (poise side down) the torque is tau = g [cos(theta) (m x - M a + m_b c) - sin(theta) (M + m + m_b) h], giving the rest angle tan(theta) = (m x - M a + m_b c) / ((M + m + m_b) h). Swings obey I theta'' = tau - gamma theta' - tf sign(theta') with I = M (a^2 + h^2) + m (x^2 + h^2) + m_b (L^2/12 + c^2 + h^2), damping gamma and knife-edge friction torque tf. About the rest angle the period is 2 pi sqrt(I / (g sqrt(S1^2 + S2^2))). A low knife edge (small h) makes the balance sensitive but slow; a high one is quick but insensitive.
Zero error and friction
If the beam's own centre of mass is not over the knife edge (offset c) the printed scale is wrong by m_b c / a; the page shows the shifted balance position. Friction makes the beam stick within a band |tan(theta)| <= tf / (g (M + m + m_b) h), which is a mass uncertainty of tf / (g a).
Checks on the page
1) A numerical integration of the equation of motion settles at the closed-form rest angle. 2) The measured period of a light-damped swing matches the formula. 3) A Coulomb-friction run stops inside the predicted stick band and a beam placed inside the band does not move.
Simplifications (honest)
The beam is a rigid uniform bar (70 cm, 250 g) with point masses hanging at beam height; no beam flexure, no knife-edge rolling, no air drag beyond a viscous term, and the beam rests on stops at 20 degrees. The load is assumed to hang vertically below its hook and does not swing.
What to try
- Drag the poise until the beam is level: the readout shows balance at x* = M a / m and the printed scale gives M.
- Double the unknown mass: the balance position doubles (graph 3 is a straight line).
- Lower the knife edge to 0.2 cm: the beam tilts much more for the same error but swings slowly (graph 4).
- Set c = +2 cm: the zero point shifts and the printed scale reads wrong by m_b c / a.
- Add friction: the beam stops short of level and the mass reading becomes uncertain.