How did the definition of complex logarithm evolve since Cotes (1682-1716)?

notes:

• This is a math-history question and Roger Cotes only used complex logarithms some time between 1682 -1716. Sir Cotes was a contemporary of Sir Isaac Newton.

• When he died, Newton is said to have said "If he had lived we would have known something."

• History of the Exponential and Logarithmic Concepts Cajori, Florian, 1939; Amer. Math. Mon. 20, 2 (Feb 1913) pp. 35-47 contains both discussion of and passages by several mathematicians of the time, including Leibnitz, John Bernoulli I, Euler, Newton and Cotes.

• I came to this after watching the Mathologer video "Euler's real identity NOT e to the i pi = -1"

From the question Difference between the formula of Roger Cotes and Euler:

Euler: $$e^{ix} = \cos x+i \sin x$$

Cotes: $$ix = \ln(\cos x + i\sin x)$$

The problem is that the complex logarithm is multivalued under the current definition. Therefore Cotes' formula is not really true anymore, but it was when he got it.

This raises the question How has the definition of the complex logarithm evolved since Cotes?

I assume the answer suggests that log was single valued in Cotes' time, since it is multivalued now. Was it in fact of the same general form as it is now but with a single breakpoint in $$\theta$$?

Was it something like $$ln(r) + i\theta$$ and $$-\pi \lt \theta \le \pi$$ for example?

If not, please explain the change in the definition of logarithm from Cotes's time to now in such a way that someone with a basic working knowledge of complex numbers who's just read Wikipedia's Complex Logarithm could understand.

below: A plot of the multi-valued imaginary part of the complex logarithm function, which shows the branches. As a complex number z goes around the origin, the imaginary part of the logarithm goes up or down. This makes the origin a branch point of the function. From here.

• $\log(ab)= \log(a)+\log(b)+2ik \pi$ Aug 18, 2017 at 15:35
• @reuns I'm asking how the definition of log has changed over time. It seems it was single-valued in Cotes' time, so that would not yet have applied then, correct?
– uhoh
Aug 18, 2017 at 15:40
• I would be skeptical that Cotes' formula is actually $ix = \ln(\cos(x) + i \sin(x))$ -- it may be a translation into modern language, and things can be lost in translation.
– user14972
Aug 18, 2017 at 15:40
• @Hurkyl I never said my math-history question was easy. I would not know where to begin, so I've asked for help here.
– uhoh
Aug 18, 2017 at 15:42
• It might be worthwhile to ask this question on the History of Science and Mathematics Stack Exchange hsm.stackexchange.com. See for example a perhaps related question hsm.stackexchange.com/questions/4907/… and some answers and comments which mention Cotes's work. Somebody there might know something about the evolution of the idea since Cotes. (And probably they would understand your question better, too.) Nov 5, 2017 at 4:55

The emphasis on the single-valuedness of a function is a fairly recent phenomenon. Cauchy in his Cours d'Analyse (1821) deals in detail with multiple-valued functions for which he uses a double-symbol notation, such as $\sqrt{~}\!\!\!\!\sqrt{x}$ (to account for both roots in this case). If so, Cotes' formula was legitimate for his historical period.