An account is presented of the history of the controversy between supporters of the crystallite and random network theories, which dominated the structural theories of network glasses for around 50 years, and is very poorly, if not inaccurately, covered in many textbooks. The development of the various theories is proposed to explain the constitution and structure of oxide glasses is outlined, starting with the early crystallite theory of 1835. The true origin of the random network theory is revealed, and a short summary is given of the arguments that led to the modern crystallite or cybotactic theory and of the reconciliation between the supporters of the (early) crystallite and random network theories. The paper concludes with a brief discussion of the present status of the theories of the constitution and structure of oxide glasses, in light of the data provided by modern experimental and computer modelling/simulation techniques.