This is what Grothendieck says -- no difference with Andreas Blass, I just thought it might interest some as additional information -- in Récoltes et semailles (p. 31/32) [my emaphasize]:
La notion de schéma est la plus naturelle, la plus "évidente" imaginable, pour englober en une notion unique la série infinie de notions de "variété" (algébrique) qu’on maniait précédemment (une telle notion pour chaque nombre premier (39)...). De plus, un seul et même "schéma" (ou "variété" nouveau style) donne naissance, pour chaque nombre premier p, à une "variété (algébrique) de caractéristique p" bien déterminée. La collection de
ces différentes variétés des différentes caractéristiques peut alors être visualisée comme une sorte d’ "éventail (infini) de variétés" (une pour chaque caractéristique). Le "schéma" est cet éventail magique, qui relie entre eux, comme autant de "branches" différentes, ses "avatars" ou "incarnations" de toutes les caractéristiques possibles.
My (poor) translation:
The notion of scheme is the most natural, the most "obvious" imaginable, to encompass in one unique notion the infinite series of notions of (algebraic) "variety", which one used before (such a notion for each prime number(39)...) Moreover, one and the same "scheme" (or "variety" of a new form) gives rise, for each prime number p, to a well-determined "(algebraic) variety of characteristic p." The collection of these different varieties of different characteristics can thus be seen as a sort of "(infinite) fan of varieties" (one for each characteristic). The "scheme" is this magic fan, which ties together, as many different "branches", its "avatars" or "incarnations" of all the possible characteristics.
End of translation. [In particular the end might be a bit messed up, as éventail also has a botanic meaning and this might be the better one with the branches, but not sure.]
Footnote 39 merely mentions that this is to include primes at infinity.
P.S. In case somebody has suggestions for improvements of the translation, I'd appreciate them.
A lot of sources mention that the adjective "tropical" is given in honor of Imre Simon, but it seems hard to find who precisely coined the term. I found some sources which attribute this to some French mathematicians. Here is what Bryan Hayes writes on the topic:
For starters, what is that word “tropical” supposed to mean? Speyer and Sturmfels explain: “The adjective tropical was coined by French mathematicians, including Jean-Eric Pin, in honor of their Brazilian colleague Imre Simon.” Pin, in a 1998 paper (.pdf), deflects the credit to another French mathematician, Dominique Perrin, again noting that the name honors “the pioneering work of our brazilian colleague and friend Imre Simon.” Simon himself, in a 1988 paper (.ps), attributes the term to yet a third French mathematician, Christian Choffrut. Apparently, no one wants to lay claim to the word, and I can’t entirely blame them. Speyer and Sturmfels go on: “There is no deeper meaning in the adjective ‘tropical’. It simply stands for the French view of Brazil.”
Best Answer
Looking at the paper of MALATIAN
"Faisceaux analytiques: étude du faisceau des rélations entre p fonctions holomorphes",
Séminaire Henry Cartan, tome 4 (1951-52), exp. n.15, p. 1-10
one finds the
Definition 3
And, in the following page:
Reading this, it seems that the original definition given by Cartan in its seminar is somehow related to the "coherent behaviour" of $\mathcal{F}$ as a subsheaf of $\mathcal{O}_U^q$, in terms of generation of the stalks.
EDIT.
However, this is not the whole story. Loking at the introduction of the book of Grauert-Remmert, as Brian suggests, it appears that the word "coherent" was actually introduced by Cartan some years before, in the middle of the '40; in fact, he investigated the so-called "coherent systems of punctual modules" when studyng the Cousin's problem. But he does not mention this previous work in his Seminar, when he introduces coherent analytic sheaves.
Grauert-Remmert write that
And Cartan himself, in its collected works, says