The study of translated texts, being the material of research, putatively equated with the hard facts of science, is only a study of variables : for each translated text represents no more than an individual effort, based mostly on intuition and governed by factors that are too varied to generalize about, hence the difficulty of formalization. Even the recent highly-rated attempt by Mona Baker to adopt a modern linguistic approach to translation from and into Arabic is based on ‘a back-translation’ approach which is far from flawless. The ‘back-translations’ are produced by the author herself and carry all the symptoms of individual bias. Even if she was a gifted translator, her ‘back translations’ would remain only ‘assumptions’ — never the ‘hard facts’ needed for a science as we understand it. Other books admit the variability and shy away from rule-setting and so come closer to ‘reflections’ on translation, with a good promise of providing a future researcher with a base for scientific work.