It is in National Semiconductor Linear Application Note 256,
"Circuitry for Inexpensive Relative Humidity Measurement", although I
was wrong about how it implemented the temperature measurement. It
just measures the forward voltage drop of one of the transistors
instead of an actual delta Vbe measurement because it does not require
absolute accuracy.

You can find it in the first application note here:

http://www.introni.it/pdf/Williams%2008%20-%20NSC%20Application%20Notes.pdf

I have seen the same temperature controlled logarithmic amplifier
trick implemented with a CA-3086 transistor array as well although it
would apply to any monolithic transistor array.

On Sat, 23 Mar 2013 17:43:38 +1100, Don Black
<donald_black@...> wrote:

>Hi David, do you have a link to the circuit using the LM389 please, I'd
>like to see how it's used.
>
>Thanks, Don Black.
>
>On 23-Mar-13 4:21 AM, David wrote:
>>
>> On Fri, 22 Mar 2013 09:41:55 -0700, Jon Kirwan
>> <jonk@... > wrote:
>>
>> I remember one clever log/antilog amplifier that used the National
>> LM389 to implement the above temperature regulation scheme. It was
>> amazingly cost effective since no temperature compensation was needed.
>> I would still be using it if the LM389 was not discontinued.