The field of systems thinking is currently fragmented, with different views and methods causing confusion. To address this crisis, a new science of systems is proposed, based on natural language and three general principles. This new approach can help with problem-solving and integrated design. The current inconsistency in how systems are viewed can be resolved by adopting this new science. The criteria for judging the quality of models in systems thinking are discussed, and the potential for new research opportunities in logic, linguistics, and mathematics is highlighted.