It is a common belief nowadays that the world economy is fairly well "integ
rated". Yet, this belief often turns out to be in contradiction with empiri
cal evidence. As a matter of fact the way distant markets interact is a que
stion that has largely been ignored by economists. In this series of two pa
pers we examine the role that space, that is to say geographical distance,
plays in the economics of commodity markets. The first of these papers pres
ents the empirical evidence while the second develops a theoretical framewo
rk. The empirical enquiry discloses several noteworthy features, e.g. (i) w
ith respect to spatial interaction there is a sharp contrast between stock
markets and commodity markets. While there is almost perfect spatial arbitr
age in the first case, this is not true for commodity markets. (ii) In spit
e of their chaotic behavior in the course of time commodity prices display
well defined spatial patterns, (iii) as in statistical physics and fluid dy
namics interactions can be described in terms of correlation length. The co
rrelation length of a set of markets is seen to increase along with the num
ber of transactions; it also increases when transport costs decline as was
the case during the "transportation revolution" of the mid-nineteenth centu
ry. Using the notion of correlation length one is able to give a quantitati
ve meaning to the otherwise ill-defined concept of market integration.