Here's "another", not claiming to be "the", answer.  In Bk. III, Marshall
defined the demand price or marginal demand price of a good as the value
of the marginal purchase. The value of the marginal purchase depends on
the marginal utility of the good and the marginal utility of money (which
Marshall generally held constant by using examples where income effects of
price changes are trivial.)  With a constant marginal utility of money
(because of the absence of the income effect), the marginal demand price
varies inversely with the quantity--not the Cournot or Walrasian approach 
in which quantity demanded varies inversely with price.  In short,
Marshall saw quantity as independent and marginal demand price as
dependent, and his axes aren't "reversed."

However, Marshall was far from consistent on this.  In his paragraph
defining the law of demand he argues:

There is then one general law of demand:--The greater the amount to be
sold, the smaller must be the price at which it is offered in order than
it may find purchasers; or, in other words, the amound demanded with a
fall in price and diminishes with a rise in price.

As I read this paragraph, it says that quantity is independent and 
marginal demand price dependent, "or, in other words," price is
independent and quantity demanded is dependent!  No wonder it's confusing.

> Interesting. Steve asked if there was a consensus view on the issue, and
> the hypothesis that seems to resonate with most of those who have piped in
> is that Marshall wanted to depict his market demand & supply curves as the
> sums of the individual curves. The source, I gather, is Blaug. But is
> there evidence for this in Marshall?
>
> Another explanation seems (at least) equally plausible, and it's the one I
> offer to my students.
>
> The answer has to do with Marshall's desire to establish doctrinal
> continuity between the new marginalist theory and Ricardo's theory. The
> classicals explained price in terms of cost of production, with cost
> understood to include a normal rate of return on capital. And they
> recognized that cost of production depends partly on the scale of
> production. In manufactures, higher levels of output can be produced at
> lower per-unit cost owing to economies of scale (see ch. 1 of The Wealth
> of Nations). In agriculture and other extractive industries, diminishing
> returns tend to prevail. The classicals up to and including Ricardo never
> conceived this commonplace empirical observation as a functional
> relationship, though in places transitional figure J. S. Mill does appear
> to treat it that way.
>
>
>
> In any case, for Marshall, who was trained as a mathematician, it was a
> small step to suppose that the relation between cost and quantity could be
> presented as a functional relationship. Following the classical reasoning,
> quantity was the independent variable: hence the supply curve, with
> (independent variable) Q on the horizontal axis. But since, in the new
> theory, the supply curve must interact with a demand curve to determine
> price, Marshall was compelled to interpret the demand function in a rather
> peculiar way in order to preserve symmetric treatment of Q as the
> independent variable: instead of asking, as we now do when we teach this
> to undergraduates, "how many units will be sold at such & such a price?",
> Marshall asks "what price can the seller obtain if he wishes to sell such
> & such a level of output?" This formulation, from which we get the concept
> of "demand price", makes quantity the independent variable. The modern
> conceptualization clearly treats quantity demanded  as a function of
> price, so the conventional positioning of the axes looks anomalous. In
> Marshall's conceptualization of the model, however, placing Q on the
> horizontal axis makes perfect sense.
>
>
>
> Early S&D theorists who had no particular interest in establishing a
> Ricardian pedigree for their models quite sensibly placed P on the
> horizontal axis & Q on the vertical axis: Thomas Wells has mentioned
> Cournot. I would add Fleming Jenkin and, indeed, Walras.
>
>
>
> Gary
>
> Gary Mongiovi, Co-Editor
> Review of Political Economy
> Economics & Finance Department
> St John's University
> Jamaica, NEW YORK 11439 (USA)
>
> Tel: +1 (990)-7380
> Email: [log in to unmask]
> ________________________________________
> From: Societies for the History of Economics [[log in to unmask]] On Behalf Of
> Humberto Barreto [[log in to unmask]]
> Sent: Thursday, February 03, 2011 10:17 AM
> To: [log in to unmask]
> Subject: Re: [SHOE] The reversed axes of supply and demand
>
>> Davenant and King were friends. Most people assume that Davenant took
>> the idea from King.
>
> Thanks, Michael.
>
> Stigler wrote:
> "A  century  later  these  calculations  were
> attributed  to  Gregory  King  by  Lord
> Lauderdale,28  without  any  known
> basis ;29  and  since  then  they  have  usu-
> ally  been  known  as  Gregory  King's
> "law."
>
> That's where I got my understanding that Davenant was the first, but I
> confused it in retelling it. Sigh.
>
> Still leaves the question of why "most people" credit King. Stigler is
> saying "without any known basis." Any thoughts or citations on that?
>
> --
> Humberto Barreto
> www.depauw.edu/learn/microexcel
>
>