Take one of the "floats" at the lower end of the temperature range, and find a way to accurately determine its mass (easy) and volume (harder), and calculate its density, noting the temperature tag that is attached. Take another float at the upper end of the temperature range and do the same. Just for grins, to ensure that your data are linear, choose a float in the middle of the range and do the same.
Plot the 3 data points, using density vs. temperature, and verify that the relationship is linear. Assuming that it is, and based on previous comments and postings, do an internet search on various hydrocarbons, and attempt to find the hydrocarbon that has the density vs. temperature relationship that you graphed. If you can't find that particular hydrocarbon, look for two or more hydrocarbons that are in the required density range, and back-calculate the mixture that is required.
I have no doubt that the data you are looking for is manufacturer's proprietary information, so you can't Google it. Thus, you either need to back-calculate it, or find a SMALL sample of it, and hire a laboratory to run an analysis on it. Good luck.
And, of course, there is an alternate method. Find a fluid that you like (assume water). Take all of the temperature indicators off of the floats. Attach new, light tags to the floats, that you can write on. Put all of the floats in the fluid of choice, and adjust the fluid temperature until only one float is neutrally buoyant and the rest are sunk. Pull that float out of the liquid and write the temperature on it. Change the temperature until the next float is neutrally buoyant, and write the correct temperature on that tag. Repeat until all floats have a number written on their tag. Admittedly, this is the "poor man's way" of getting the job done, but it is probably the most direct way.