In this paper we take position in the ‘citation theory’ debate. First we revisit relevant earlier work of our group and try
to assemble the findings. We criticise the constructivist fashion in sociology of science concerning citation practices. With
statistical arguments we show the strong limitations of any ‘citation theory’ at the ‘citer side’. We emphasize that citations
should be conceived of as ‘binding properties’ of an individual publication, from which many types of structuring follow.
As keywords also have such binding properties at the same time, and as there are empirically established relations between
the citation domain and the word domain, it is useless to develop a model concerning citations only. We envisage an interesting
development, both theoretically and empirically, of what we would like to call ‘bibliometric chemistry’.