SUMMARY
All through the literatura, the question about what is a logical constant has recieved many answers, from model-theoretic aproaches (Tarski; 1966), (Sher; 1991), (Bonnay; 2007) to answers that focus in the inferential practice as meaning (Dummett; 1991), (Prawitz; 1965), (Lorenzen; 1955). Detractors of the second tradition presented many ineludible incovenients, in particular, the logical constant named ?tonk? (Prior; 1960). Inferentialist tryed many solutions, in particular they presented the concept of ?harmony?. The goal of this paper is to show that the different criteria of ?harmony? used in the proof-theoretic semantics to determine what is and what is not a logical constant fail to be necessary or sufficient. I will show the philosophical reasons that make this concept appear and then i will describe the different ways in wich the literatura understads the concept of ?harmony?. Then I will show that they subgenerate or overgenerate connectives with some counterexamples. Finaly, I will explain some philosophical reasons that should delimitate where to go towards a satisfactory definition of ?harmony?.