Cauchy got to a rigorous foundation of calculus by rigorously defining infinitesimals out of existence! His infinitesimal-free foundation for Calculus spearheaded a new rigour in mathematics, and was so triumphant that, a century later, Russell announced that infinitesimals had been an absurd notion all along. To this day, infinitesimals play absolutely no formal role in standard real or complex analysis. Instead, we talk exclusively in terms of limits, an idea heavily promoted by Euler, but only formally defined by Cauchy, and only in the most elementary terms of real numbers. Infinitesimals make no appearance at all (I'm happy to elaborate).
Russell's declarations of absurdity were completely wrong, however, and infinitesimals were shown to be not only logically coherent, but a fairly natural logical notion discovered first by Robinson with the hyperreals, and given an alternative formulation be Conway with the surreal numbers. You can do analysis in the hyperreals, and my old boss did his doctoral research near enough successfully vindicating Newton's calculus as being rigorous if you interpret him as working in the hyperreals.
However, pretty much all maths students today do the standard analysis that was laid down by Cauchy (with final gaps filled in by Dedekind and Cantor), and never see an infinitesimal anywhere. This makes sense, since the construction of the hyperreals ends up building on top of the standard reals anyway.
Thank you for you comment, which brings a bit of sanity in this thread.
And I take you point.
I did two years of maths and physics a very, very long time ago in Paris and then had to drop out, there and then, without finishing the cursus. So, I'm not going to pretend I'm the expert around here on this issue.
Still, I am in effect discovering now that this had been, historically, something of a controversial issue. To me it's never been and still isn't. I take the term 'infinitesimal' to have a precise meaning, i.e. to refer to a clear concept, one that I thought most mathematicians were confortable with. So, I accept the fact that it's a controversial issue but not through and through. Which now I have to explain.
Bear in mind that you may be more interested in the technically mathematical aspect of this issue while my interest is more in the conceptual landscape behind it.
Personally, I make a clear distinction between concepts of the mathematical world and how we mathematically define the properties we associate with these concepts. For example, we have a concept of infinitesimal built around the concept of continuous curve around a particular point called the 'limit'. Broadly, this concept is known and understood through a simple diagram showing, say, two axes, a curve representing the function itself, curve which will be continuous near one specific point. We don't really need any formal definition to have clear enough ideas about the concept of infinitesimal: given that the function is continuous, we can choose a point on the curve which will be as close as we may wish from the point we call the limit. If someone selected one point, somebody will always be able to find and select another point even closer to the limit. Intuitively, there's no other limiting point than the pre-defined limit itself. So that's one thing.
The second thing is how we're going to express those lovely intuitive ideas in a formal and rigorous way so that we can make progress in discovering the properties of our concepts that may be not so obvious and about which we should expect to have recurring disagreements if we were to discuss them informally.
According to this distinction, I don't believe for a moment that any of the mathematicians you've discussed here with beero had any issue with this notion of infinitesimal. I take the historical debate between them, and probably many others, to have been about the second aspect: how do we formally specify our mathematical concepts in such a way as to preclude any impasse and disagreement. Formal definitions are notoriously difficult to articulate straight away. It's an iterative process and we always need several people to bring a fresh perspective on the discussions.
No doubt, some more metaphysical points would have clouded the discussions. The idea of infinity is definitely difficult to formalise. Still, I am convinced that all these people mostly had the same basic conception in mind. It was therefore only a matter of agreeing on the language to use to talk about it to reduce any unnecessary friction.
Accordingly, the particular language agreed by mathematicians after some huffing and puffing should be understood as just the formal way retained for talking about the original conception, which won't have changed in the process because it has been unproblematic from the start. I would even go so far as saying that the concept of infinitesimal is not only easy for mathematicians but it is also intuitive for most people, even though few people would be able to articulate any formal definition of it.
Consequently, I don't see the language eventually retained by mathematicians for expressing the properties associated with continuity of a curve in the neighbourhood of a point as primary. What is primary is our intuitive concept of a curve being continuous near a limit.
In other words, while Cauchy's definition doesn't mention infinitesimals, you really understand what this definition means if you understand the concept of infinitesimal. You need the formal definition to go beyond this and make progress.
I wish I could have understood this at the time.
EB