express the facts with absolute simplicity, without any moulding, alteration or addition, in whose elaboration every science consists, and which precede all knowing, every judgment regarding the world. It makes no sense to speak of uncertain facts. Only assertions, only our knowledge can be uncertain. If we succeed therefore in expressing the raw facts in "protocol statements," without any contamination, these appear to be the absolutely indubitable starting points of all knowledge. They are, to be sure, again abandoned the moment one goes over to statements which are actually of use in life or science (such a transition appears to be that from "singular" to "universal" statements), but they constitute nevertheless the firm basis to which all our cognitions owe whatever validity they may possess. 

Moreover, it makes no difference whether or not these so-called protocol statements have ever actually been made, that is, actually uttered, written down or even only explicitly "thought"; it is required only that one know what statements form the basis for the notations which are actually made, and that these statements be at all times reconstructible. If for example an investigator makes a note, "Under such and such conditions the pointer stands at 10.5," he knows that this means "two black lines coincide," and that the words "under such and such conditions" (which we here imagine to be specified) are likewise to be resolved into definite protocol statements which, if he wished, he could 
in principle formulate exactly, although perhaps with difficulty. 

It is clear, and is so far as I know disputed by no one, that knowledge in life and science in some sense begins with confirmation of facts, and that the "protocol statements" in which this occurs stand in the same sense at the beginning of science. What is this sense? Is "beginning" to be understood in the temporal or logical sense? 

Here we already find much confusion and oscillation. If I said above that it is not important whether the decisive statements have been actually made or uttered, this means evidently that they need not stand at the beginning temporally, but can be arrived at later just as well if need be. The necessity for formulating them would arise when one wished to make clear to oneself the meaning of the statement that one had actually written down. Is the reference to protocol statements then to be understood in the logical sense? In that event they would be distinguished by definite logical properties, by their structure, their position in the system of science, and one would be confronted with the task