TY - JOUR T1 - The Strehler-Mildvan correlation is nothing but a fitting artifact JF - bioRxiv DO - 10.1101/064477 SP - 064477 AU - Andrei E. Tarkhov AU - Leonid I. Menshikov AU - Peter O. Fedichev Y1 - 2016/01/01 UR - http://biorxiv.org/content/early/2016/07/18/064477.abstract N2 - Gompertz empirical law of mortality is often used to parametrize survival fraction as a function of age with the help of mere two parameters: the Initial Mortality Rate (IMR) and the Gompertz exponent slope, inversely related to the Mortality Rate Doubling Time (MRDT). Furthermore, the two parameters are believed to be related through the Strehler-Mildvan (SM) correlation, which suggests that the more animals die during adolescence, the longer is the lifespan of the survived individuals. Even though there have been quite a few doubts expressed against the very existence of the correlation, it is still widely believed that the theory of ageing and mortality behind the SM correlation provides a mechanism-based explanation of Gompertz's law. In this Letter, we concentrate on uncertainties of identification of the Gompertz parameters by fitting from experimental survival records. We show, that whenever the number of animals in the experimental cohorts is insufficiently large, the use of least-squares fitting is rather a convenience for data visualization, than a robust procedure to amalgamate and smooth discrete data. We present an analytical explanation behind specific difficulties accompanying the fit once the average lifespan of the species exceeds MRDT. We demonstrate, that under such conditions a fit to the Gompertz law becomes unstable, and fails to produce a unique combination of the demographic parameters. In fact, one gets the whole degenerate manifold of the Gompertz parameters, which is nothing else but the line, corresponding to the proper approximate value of the average lifespan, and, at the same time, coincides with the SM correlation. We show that the employment of the age-independent Makeham mortality term does not resolve the degeneracy problem. Therefore, we have to conclude, that the average lifespan persists as the only stable feature, which can be reliably inferred from survival statistics in an experiment with a finite number of animals. The SM correlation, in this case, is nothing but an iso-average-lifespan degenerate manifold on the IMR versus \alpha plane, is thus a fitting artifact, and its observation in a data set is an indication of insufficient statistical power, rather than a biological reality. ER -