[en] The transgenerational effects debate of recent years can, in the light of current knowledge, be seen as a working example of the operation of a powerful range of errors, biases and confounders. These have often helped to obfuscate the issues addressed to the disbenefit of interested parties such as vicinity populations and workforces. The possibility of transgenerational effects has been entertained as a theoretical constant throughout the history of radiation science being given particular direction and focus by the work of Mueller in the 1920's. The absence of such effect in bomb survivors was therefore somewhat surprising to researchers even though this relative radio resistance was confirmed by later animal studies, such as by Russell and Selby. For emotive and situational reasons the renewed transgenerational debate of the last couple of decades has focused largely on childhood leukaemia, a very late, even remote-manifesting putative, transgenerational effect. This effect has now been demonstrated to be mainly due to confounding, most likely by population mixing, even in the sentinel study population of Seascale, near Sellafield. Little attention had been paid to the nature of the biological plausibility of putative transgenerational effects of ionizing radiation in terms of likelihood and closeness of fit. Thus there is a likelihood gradient of expected magnitude of effect which may be predicted to run from early to late manifesting defects. This would be expected to be high for pre-implantation loss and low for stillbirths or childhood cancer. Cited biological concordance also seldom takes regard of dose and dose rate. This is a particular problem because many epidemiological studies use more or less crude surrogate of dose such as monitored/never monitored or mechanical proportionalisation of annual dose summaries to shorter critical periods (such as spermatogenesis). In questionnaire studies, which are often regarded as inescapable in reproductive epidemiology, strong ascertainment bias may be identifiable. There is also uncertainty about the limitations of using certain types of controls (e.g. nuclear workers vs. non-nuclear workers) in case-control studies because of concerns over selection and overclassification. A decline is discernible in the modern hypothesis base in this field which has thus moved from, 'x=effect' to 'x + y=effect' (where x is ionizing radiation and y is one or more unknown factors). Whilst this latter type of conditional reasoning may be seen as arguable in terms of multistage phenomena, it also falls into a fundamental, scientific, philosophical trap known as 'undetermination of theory by data' (Duhem-Quine thesis). (author)