The idea that CO2 in the atmosphere could cause global warming was first proposed near the beginning of the 20th Century, by Svante Arrhenius. He thought that a doubling of CO2 could cause a rise in global temperature by 5 C. And calculated that average global temperature was 15 C. And that CO2 levels could explain Ice ages.

http://www.lenntech.com/greenhouse-effect/global-warming-history.htm

Today it is often said that average global temperature is about 15 C. But what this means is not precisely defined. For instance, does mean temperature adjusted to some elevation or are we talking about near ground temperature reading taken uniformly around the world?
The latter appears to be the way it's done and obviously there are not currently weather stations uniformly spread around the world, so some kind of adjustments are made.

But perhaps, more interesting is how Svante Arrhenius arrive at his estimate of global average temperature. Though the Antarctic Peninsula was first seen in 1820 and 1821 was first time anyone set foot anywhere on the continent not much was know about it until after Svante Arrhenius was dead. See timeline:
http://www.south-pole.com/p0000052.htm

How could Svante Arrhenius know that the Antarctic was 14,200,000 km² and it's average temperature was around -50 C. Here was a continent bigger than any other country except Russian [17,098,242 km²] and almost twice the size of Australia [7,692,024 km²].

And for that matter, what would Svante Arrhenius even know about Russia's average temperature? Btw, does anyone happen to know now what Russia average temperature is- other than on average it's below freezing and generally colder than Canada, with 40% of land area having permafrost. Somewhere around freezing is about the best I could come up with.
Now perhaps, Svante Arrhenius simply got average temperatures from ships traveling the oceans. After all the ocean do cover most of the Earth surface area and if guessed correctly about the average ocean temperature, the land temperature could be more or less be
ignored.

Now according to this chart:

Svante estimate of 15 C was wrong, as it should been according to graph about 14 C or around 13 1/2 C, though it's possible the graph is wrong & Svante was correct or possibly neither are correct.
Also it seems that the 5 C rise from a doubling of CO2 is incorrect. If the CO2 levels at Svante's time were somewhere around 300 ppm, then one would assume that 300 ppm level would be adding 5 C or more to the global temperature and therefore a doubling would be adding an additional 5 C.
I don't anyone today thinks that even at the present level of around 380 ppm of CO2 that this amount it is adding 5 C to global temperatures, but rather it's usually thought that 380 ppm is adding somewhere around 2 C.

Another aspect of this is that I don't believe that Global CO2 level have ever been much lower than 180 ppm of CO2. This is simply due to the fact that plants need at minimal level around 200 ppm of CO2 and probably can't survive at say 160 ppm and if plants can't survive all animals will die. In addition the ice core records don't show levels lower than about 180 ppm. And these ice core records show a range of about 180
to around 300 ppm of CO2- a difference of at most of 120 ppm.
So, even with Svante's 5 C per 300 ppm that only gets a 2 C range in temperature, and the global temperature difference between Ice ages and interglacial is generally more than 10 C. And if instead you allow only 2 C per 300 ppm, then it's only causing a .8 C increase.

Of course more important than this is the fact that rising CO2 level follow rising global temperatures instead causing global warming, but even the imagined role of CO2 reinforcing or maintaining warming trend it only amount at best to .8 C.
Then of course looking at the graphs one see that with the higher levels of CO2, global temperatures then drop and therefore at it's most *potent* levels is easily overwhelmed by other factors affected global climate.

So, it seems to me that any way you look at it, CO2 can only play at best a very minor role in ice ages and interglacial periods. And I believe rather than 2 C per 300 ppm, it's more likely and more scientifically accepted that it's around 1 C per 300 ppm increase of CO2.

According to and same place of above graph:
http://www.earth-policy.org/Indicators/Temp/2008.htm

Earth average temperature in 2007 was 14.73 degrees C and "northern hemisphere alone, 2007 temperatures averaged 15.04 degrees C".

It doesn't make much sense to me that the northern hemisphere is warmer than global average temperatures, because the Northern hemisphere is less dominated by ocean areas- even though Antarctic must put serious dent in the Southern Hemisphere average temperatures.

In terms of all land areas on Earth the country of Russia is 11.5% of land area. And Canada has 6.7%. The US has 6.5% and the US average temperature is about 12.3 C: "The average U.S. temperature for 2007 was 54.2°F; 1.4°F warmer than the 20th century mean of 52.8°F".
http://www.noaanews.noaa.gov/stories2008/20080115_warmest.html

No one should imagine that Canada is as warm in terms of average temperature compared to the US. Russia is colder than Canada and European continent [the EU {27} equals 3% of total land area] is as cold or colder than the US.
I roughly figured that Mongolia [1.1%] average temperature is about -1.5 C. It's north of China and south of much of Russia. China [6.5%] including Tibet; is probably as cold or colder than the the US.

As rough rule probably all land areas above the 40th parallel are cooler than the US. And as wild guess any oceans above the 50th parallel are cooler than the US. The average temperate of Mexico City [elevation 2,300 meters] is 15 C and it's slightly below the 20th parallel, though obviously the country of Mexico has higher average temperature than than the US- probably 15 C or higher.

In terms of broad areas: in Northern Hemisphere probably coldest large area is Greenland and it's average temperate is less than -10 C and has area of 2,166,086 km² 1.5% of land area of Earth.
The Arctic Circle [21,000,000 km²] which includes parts of Canada, Russia, Greenland, the polar sea ice, with some open ocean year round, etc. Average temperature is probably less than -5 C. Russia is probably at or less than 0 C. Canada is probably average is less than 5 C.

Trivia bonus points:
Highest average annual mean temperature (world): Dallol, Ethiopia (Oct. 1960–Dec. 1966), 94° F (34.4° C). (U.S.): Key West, Fla. (30-year normal), 78.2° F (25.7° C).
http://www.infoplease.com/ipa/A0001431.html
"The climate of Ethiopia varies mainly according to elevation."
http://encarta.msn.com/encyclopedia_761573854/ethiopia.html
Below 6,000 ft: 27°C And about 6,000 and 8,000 ft: 22°C

I was going to say that the hottest average temperature wouldn't get over 30 C, but I thought check this, and I'm glad did as I would have wrong. But in large regions the average temperature probably doesn't gets above 30 C. Though there probably large tropical regions which around 28 C. And it's these large regions particularly huge ocean regions which will increase the average temperature of Northern Hemisphere.
Northern Hemisphere is half of surface area, so it's 255,000,000 km²
About 100 million sq kilometers is in tropical region [between the equator and Tropic of Cancer]. The area is mostly ocean, and includes about 1/2 of Mexico and all of Central America, a chunk out of the African Continent, part of Saudi Arabia, India, and most of South East Asia. Ethiopia is in this zone and roughly is 25 C. And Philippines also with this region: "The average yearly temperature is around 26.5 °C (79.7 °F)"
http://en.wikipedia.org/wiki/Philippines
And Taiwan which cut in half by Tropic of Cancer is about 24 C in it's southern half

So, let's just say this whole region averages around 26 C and it is about 39 % of northern Hemisphere. Greenland is 2,166,086 sq km and region would hold 49 Greenland size land
areas. If you average a 2,166,086 sq km part of tropics with Greenland you get an average temperature of 26 + -10 = 8 C. And you take the combined 4,332,172 and add to another 4,332,172 piece of tropics, and you get 8,664,344 area with average temperature of 17 C. So Greenland would get "borged" into warmer area than global average temperature by less than 1/10th of the Tropical region area.

For fun and to understand scale, let's do the same thing with the Southern Hemisphere. We will assume tropics of southern Hemisphere is also has average temperature of 26 C. The Antarctic is 14,200,000 km² and about -50 C. The southern tropics would hold about 7 Antarctic size areas.
So you take 14,200,000 km² of tropics and you get the average: 26 + -50 = -12 C. So you have an area of 28,400,000 km² at average temperature of -12 C. You repeat by getting a different 28,400,000 km² piece of tropics: 26 + -12 = 6 C. So now you have an area 56,800,000 km² at average 6 C facing off the remaining tropical area of 57,400,000 sq km. So you repeat by adding 56,800,000 km² of tropics and get 26 + 6 = an average temperature of 15 C. So the tropics could barely borge the Antarctic and if the Antarctic was actually a little cooler or tropics a little cooler the colder than 15 C wins the day. And if every south of the 50th parallel were engaged in the battle the cool ones will easily win.

Anyhow, it's been a long post.