In the Nirenberg and Matthaei experiment the artificial mRNA, polyU, was translated into polyphenylalanine in a cell-free system, establishing that UUU was the codon for Phe. How did this work as the artificial mRNA lacked an AUG initiation codon, which we now know is required for translation of cellular mRNAs?
This is an intelligent question that highlights the ‘sleight of hand’ simplifications employed in many text books to make experimental science appear cleaner and more logical than is in fact the case. From what we now know about the requirement for an AUG initiation codon in protein biosynthesis these experiments should not have worked.
The reason they did was that they were performed at an unphysiologically high concentration of Mg2+ which overcomes the need for specific initiation (and, incidentally, allows other reactions of protein synthesis to proceed in an abnormal manner, particularly the codon-directed tRNA binding reaction used to finish the deciphering of the genetic code). You should also bear in mind that the efficiency of the reaction was probably much less than that within the cell, but this was of no concern as all that was required was that a product be synthesised.
I worked in this area in the late 60s/early 70s and the importance of magnesium concentration was just common knowledge in the field, so it is difficult to cite a reference to support my assertion. However a brief search brought up a contemporay paper on the effect of magnesium on initiation of protein synthesis in phage RNA, illustrating its importance. There are, no doubt, others.