Number synthesis of kinematic chains usually involves the generation o
f a complete list of kinematic chains followed by a time-consuming, co
mputer-intensive procedure for the elimination of isomorphs. A signifi
cant unsolved problem in number synthesis is the guaranteed precise el
imination of all isomorphs. Since there is no efficient algorithm for
always determining whether two kinematic chains are isomorphic, any ''
efficient'' algorithm has a finite probability of rejecting a unique,
potentially useful, chain. This paper reviews the history of number sy
nthesis and presents a new orderly method for synthesising kinematic c
hains. This new Melbourne method guarantees to produce a complete list
of chains, which, only when some doubt to the uniqueness of a chain e
xists, may include an isomorphic chain. As a consequence, this techniq
ue produces significantly fewer isomorphs in the output list than do p
revious techniques; often no isomorphs are produced by the method what
soever. It is proposed that in many situations where the synthesis of
kinematic chains is required, the processing of duplicate chains in th
e early stage of design is preferable to the omission of a potentially
useful chain.