Lavoisier Paper

Lavoisier Paper

Today my paper on the statistics of global average temperature was posted on the Lavoisier Group’s web site:

It is intended as a popular account of my paper in Energy and Environment which recently appeared on-line. Unlike that paper, it includes no equations or mathematical symbols. It examines the old-fashioned deterministic world-view of the applied mathematicians who run the climate models and compares it to the 20th century idea of “stochastic process” which more readily accommodates the scientific method.

The underlying assumption of the  stochastic approach is that every state is dependent not on the time per se but only on previous states by a process known as auto-regression. There is also assumed to be an additional, unknown, random component called “the innovation”. This statistical approach allows the use of well established statistical methods to test for drifts and cycles in the data.

The conclusion? There is no significant trend in global average temperature and therefore no need to look for causes. At time scales of less than a millennium global temperature variations are just red noise.


Reid,J. (2017) There is no significant trend in global average temperature. Energy and Environment 28, 3, 302–315.

A copy may be downloaded here.

3 Replies to “Lavoisier Paper”

  1. Thank you John. This is important work. The deterministic approach is useful in closed systems that rarely actually exist but maybe useful to explain something in isolation. The stochastic approach is how we need to deal with the real world as observed where the system is open and never fully knowable. We are very prone to seeing patterns where none exist. Your work is a wonderful argument highlighting the fact. Many sophisticated looking models go off half cocked for failing to recognise they are fully open systems and by definition indeterminate to some degree.

    1. To me it’s not about “open” and “closed”, it’s about deterministic and stochastic. The important thing is that differential equations are deterministic and turbulence in fluids is stochastic. Furthermore all of the differential equations of fluid dynamics only apply to a continuum whereas no real fluid is a continuum because of the discrete nature of matter. Hence the application of differential equations to fluid dynamics is not justified. Some statistical process must be found which relates to existing formulations in a similar way that statistical physics relates to thermodynamics.

      I agree about seeing patterns where none exist. It used to be called “superstition”.

      1. Thank you for the reply John, point taken. I’d gone off on a tangent. It had been a while since I read your earlier paper on this. I enjoy the thought processes you describe.
        Kind regards,

Comments are closed.