The second law of thermodynamics says that the entropy – or disorder – of an isolated system undergoing a cyclic process will increase or remain the same. But this law only applies to large systems over significant periods of time. To explain the behaviour of smaller systems, Evans and colleagues devised their ‘fluctuation theorem’, which calculates the probability that entropy will be consumed at any point in the cycle. They found that it predicted that measurable violations of the second law would take place in small systems over short time-scales.

To test the idea, the researchers put about 100 latex beads – each 6.3 µm across – into a water-filled cell, which was placed on the stage of a microscope. The researchers focused a laser onto one of the beads, which induced a dipole moment in the bead and drew it towards the most intense region of the electric field in the laser beam. The force acting on the particle near the laser focus was harmonic.

With the bead trapped, the researchers moved the microscope stage backwards and forwards repeatedly, dragging the bead in and out of the laser focus. The stage moved through 540 such cycles in ten seconds, and the team measured the position of the bead a thousand times every second. Combining these measurements with the laser power and fluid drag, Evans’ team was able to calculate the forces acting on the bead – and its entropy production – as it moved.

Evans and co-workers found that – during some trajectories – entropy was consumed rather than generated. This effect was seen when the researchers looked at the bead’s behaviour over periods of about a tenth of a second. Over periods approaching two seconds, the proportion of entropy-consuming trajectories fell, and above two seconds, none were observed. The team also found that their results closely fitted a computer simulation of the fluctuation theorem.

Evans and colleagues say that their discovery could be important in the design of nanomachines. They also point out that as thermodynamic systems become smaller, the probability that they will run ‘in reverse’ increases, and this could improve our understanding of how many small biological systems – such as ‘protein motors’ – work.