In engineering design, form has traditionally been specified explicitly using blueprints. This thesis explores an alternate way of specifying form built on interactions between dynamical systems. This alternate form specification is based on ideas from natural development. Inspired by termite nest building behavior, dynamic developmental systems are proposed as an alternate method to produce and represent structure designs, which when compared to the conventional blueprint method are a more robust form specification method, more adaptive, and even able to self-repair. Developmental systems are uses here as a method of form specification and an evolutionary algorithm is the method of design chosen to explore the capabilities of these developmental systems. Evolutionary algorithms have already been widely studied and proven to be an effective method of finding solutions to tough problems, and in this work they are simply a validated tool being used. The experiments included in this work use developmental systems with high degrees of system-environment interaction and show the importance of a subtle and often overlooked difference between two similar kinds of systems. An important distinction is being made between systems which both use feedback from the environment. These systems are referred to as the reactive system and the interactive system. The reactive systems simply use environment feedback during their development, whereas the interactive systems not only use environmental feedback but actually form a two-way dynamic feedback cycle WITH the environment. Our control experiments are the systems with one-way feedback which have a system-environment interaction level where the system uses information from the environment during its development but does not affect the environment?s dynamics. Our experiment systems ii use dynamic feedback, which allows them to affect the dynamics of the environment while simultaneously the environment reacts to this stimulus, forming a two-way feedback loop which makes the system more situated in the environment. The experiments in this thesis used the evolutionary algorithms to search for systems which fulfilled the desired effect on the environment. In this case this effect is to build a structure that causes the average temperature in the environment to come as close as possible to a target temperature, which is specified at the beginning of the evolutionary run. Both types of systems were evolved using evolutionary algorithms and those systems which used dynamic environmental feedback consistently displayed better performance.