The broadcast language is a programming formalism devised by Holland in 1975, which aims at allowing Genetic Algorithms (GAs) to use an adaptable representation. A GA may provide an efficient method for adaption but still depends on the efficiency of the fitness function used. During long-term evolution, this efficiency could be limited by the fixed representation used by the GA to encode the problem. When a fitness function is very complex, it is desirable to adapt the problem representation employed by the fitness function. By adapting the representation, the broadcast language may overcome the deficiencies caused by fixed problem representation in GAs. This report describes an initial detailed specification and implementation of the broadcast language. Our first motivation is the fact that there is currently no published implementation of broadcast systems (broadcast language-based systems) available. Despite Holland presented the broadcast language in his book “Adaptation in Natural and Artificial systems”, he did not support this approach with experimental studies. Our second motivation is the affirmation made by Holland that broadcast systems could model biochemical networks. Indeed Holland also described how the broadcast language could provide a straightforward representation to a variety of biochemical networks (Genetic Regulatory Networks, Neural Networks, Immune system etc). As these biochemical models share many similarities with Cell Signaling Networks (CSNs), broadcast systems may also be considered to model CSNs. One of our goals, within the ESIGNET project, is to develop an evolutionary system to realize and evolve CSNs in Silico. Examining the broadcast language may provide us with valuable insights to the development of such a system. In this paper, we initially review the Holland broadcast language, we then propose a specification and implementation of the language which is later illustrated with an experiment: modeling different chemical reactions.