We introduce the possibility of research duplication under cooperative R&D and technology transfer under non-cooperative R&D in Marjit (1991, Eco. Lett.) model and derive some new results over the existing literature. In the analysis the R&D cost, the success probability and the size of the innovation, all these play a crucial role. For instance, when the probability of successis in the intermediate range, contrary to Marjit (1991), non-cooperative R&D is preferred to cooperative R&D provided the size of innovation is large, otherwise cooperative R&D occurs. On the other hand, the possibility of royalty licensing enhances incentives for non-cooperativeR&D.