To tout its Olympics ratings success, NBC is borrowing a measurement from the online world: cumulative audience.

Akin to unique visitors to a Web site, NBC’s favorite number these days is a count of the number of people who have tuned in at some point to U.S. television coverage of the Beijing Games. Through Tuesday, that number stood at 203 million — four million more than the count at this point during the 1996 Olympics in Atlanta, and six million shy of the record set that same year. The total count already has exceeded viewership in 2004 in Athens, by a margin of fewer than one million.

But it’s unclear why television advertisers would care much about how many people tune in at some point over the 17 days of coverage on NBC networks. On the Web, extended advertising campaigns can eventually reach much of a site’s monthly audience. On TV, an advertiser gets its message in front of people only if they’re tuned in when it airs.

“They care about the cumulative audience,” Patricia McDonough, senior vice president for insights and analysis for Nielsen Corp.’s Nielsen Media Research, told me about advertisers, “but not as much as they care about their minute of advertising.”

That said, both she and NBC Universal spokesman Brian Walker said advertisers care about mass appeal. “Every advertiser wants to be associated with an event that has become one of the most viewed in television history,” Mr. Walker said.

The quest for ratings records is an arduous one. Networks often emphasize that comparisons with TV history are tough for today’s ratings numbers because of the wider array of competition, on screen and online.

But there are also more potential viewers, because of population growth. The comparison between Beijing and Atlanta is a bit unfair to those 1996 Games: There were fewer than 270 million Americans in the summer of 1996, compared to 305 million now, according to U.S. Census Bureau estimates.

By the measure of how many people didn’t watch, Beijing surpasses Atlanta’s pace by 31 million (35 million more Americans now, minus the four million more who’ve tuned in), and is level with the pace for Athens, when there were 12 million fewer Americans. And it’s tough getting those stragglers, whose overall interest in the Olympics is questionable. Between the second and third day, NBC’s networks added 31 million new viewers. On Day Six, total audience climbed by just six million. And between Monday and Tuesday, the number rose by just three million.

Of course, all of these numbers are subject to my usual cautionary notes about Nielsen’s methodology, but those methods have their advantages. For total audience, Nielsen tracks its panel of users to see whether they have tuned in at some point; because the company is tracking the same people over time, it can account for duplicate viewers. Today’s numbers “should be more reliable” than those in 1996, Ms. McDonough said.

Further reading: NBC is using other unusual numbers to track Olympics viewership. An estimate of world-wide audience size for an Olympics basketball game appears overblown. There’s news, features and photos from Beijing on the Journal’s Olympics page.

It was fitting that the arcane tiebreaker rules for Olympics gymnastics awarded China’s He Kexin with a gold medal in the uneven bars and left American Nastia Liukin with the silver. That’s how the Olympic medal count has gone: The U.S. has kept just ahead of China in the overall medal count, but China has blasted ahead in golds.

Gold medalist Lu Chunlong of China poses on the podium during the award ceremony of the men’s trampoline Tuesday. (Associated Press photo)

Through Tuesday’s events, China had 43 golds, 57% of its 76 total medals. A typical country has 33% golds among its medal haul, of course, because for each event there’s a gold, silver and bronze for the taking. And the U.S. is in line with that rule of thumb, with 26 golds out of 79 total medals.

How unusual is the predominance of gold in China’s medal metal? That’s what Numbers Guy reader Doug Morgan wanted to know. The answer: Quite unusual.

No country that led the gold-medal count has matched China’s golden percentage of 57%, according to my analysis of historical summer Olympics results from databaseOlympics. Just three times has this percentage topped 50% for the gold leader: The U.S. in 1912, and the Soviet Union in 1952 and 1972. No country has won more than two medals without winning a bronze or silver (10 times countries have won just two golds, most recently the Bahamas and Slovenia in 2000). Norway in 2004 had the highest percentage of golds with more than two medals: five golds out of six medals. And just two countries with 10 or more gold medals had a higher percentage of golds than China has this year: Hungary in 1936 (10 golds out of 16 total medals) and Spain in 1992 (13 of 22).

China is hosting the Games this year, as did Spain in 1992. But overall, there doesn’t seem to be a noticeable host effect on the percentage of golds. On average, 35% of host nations’ medals have been gold. At the opposite extreme from China this year, host Canada in 1976 won 11 medals, none of them gold. (That wasn’t the most surprising gold shutout; in Finland in 1952, Germany won 24 medals, none of them gold.)

China’s gold rush aligns with the country’s — and much of the world’s — preferred measure of Olympics dominance: The gold-medal count. And the U.S. can claim that it is on top at these Games, because it leads in total medals. If this holds up, the medal count will have a split title for just the fourth time in history. Ironically, it was the U.S. that led in golds but not medals — its preferred measure — in each of those three earlier Olympics (in 1896, 1912 and 1964).

Now that Michael Phelps has his record eight gold medals at a single Olympiad, here’s a quick update of Friday’s post comparing Mr. Phelps’s results in the last week with Mark Spitz’s at the 1972 Olympics. Though Mr. Phelps exceeded Mr. Spitz’s medal count, by other numerical measures he was less dominant compared to his peers, a contrast made more marked by Mr. Phelps’s razor-thin margin in the 100-meter butterfly on Saturday.

Michael Phelps with one of his octet of golds (Associated Press Photo)

Number of medals: Mr. Phelps exceeded Mr. Spitz’s seven by one, with two more men’s swimming golds available. He matched Mr. Spitz’s mark of seven world records.

Head to head: In their six common events, Mr. Spitz won by an average margin of 1.58%, compared to just 0.67% for Mr. Phelps. Mr. Spitz set world records in all six, by an average margin of 0.93% compared to the pre-Olympic records. Mr. Phelps set new marks in five of the six, for an average margin 0.75% faster than the previous standard.

Across all their events, Mr. Spitz’s average winning margin was 1.47%, compared to 0.86% for Mr. Phelps. Mr. Phelps’s new records exceeded the old by 0.7%, compared to 0.87% for Mr. Spitz.

Individual events: The above analysis includes relay events, which depend on three other swimmers. Each champion earned gold in three of these. Discard these, and Mr. Spitz has an edge in winning margin (1.44% to 1.06%) and record margin (0.58% to 0.33%).

Context: Though Beijing’s Water Cube has earned raves from swimmers and facilitated spates of new records, Munich in 1972 featured even friendlier waters. There were new records in six of eight events Mr. Spitz skipped, compared to just four of the eight completed events Mr. Phelps skipped. (The 10-kilometer freestyle is still to come.)

Mr. Phelps’s times beat the old records by an average of 0.7%. In the other eight completed swimming races, including those that didn’t feature new records, the winning time was an average of just 0.07% better than the pre-Olympics record. Mr. Spitz’s records beat the old ones by an average of 0.87%; his male counterparts beat record times by an average of 0.34%.

Summary: Mr. Spitz was slightly more dominant in his races, set a new world record in each event and outdistanced prior record holders by a wide margin. However, Mr. Phelps’s records have been slightly more impressive compared to other male winners at the same Olympics.

If Michael Phelps wins his remaining two events, he’ll have eight gold medals, more than any other athlete at a single Olympics. By common consensus, that will make his feat the best single Olympics ever. But a closer look at the numbers suggests Mr. Phelps has some work to do to match Mark Spitz’s seven gold medals in swimming at the 1972 Olympics. Here’s a breakdown of how the two U.S. swimmers compare through Friday’s events in Beijing.

How do U.S. Olympics swimmers Michael Phelps (left) and Mark Spitz compare? (AFP/AFP/Getty Images)

Number of medals: If Mr. Phelps wins out, he’ll have one more gold medal this year than Mr. Spitz in 1972, but each will have won just under half the total events. That’s because there were 15 men’s events in 1972, but two more this time around: the 50-meter freestyle — which Mr. Spitz has said he would have won if given the chance — and the 10-kilometer marathon race. And Mr. Spitz set a world record in all seven events; Mr. Phelps has done the same in his first six. (Incidentally, by available medals neither swimmer can compare to Olympians who sweep in sports with sparser hardware. The plethora of swimming medals makes comparisons between sports a bit unfair.)

Then again, Mr. Phelps isn’t swimming in either of these events. In fact, he’s swimming six of Mr. Spitz’s events, plus two that were available in Mr. Spitz’s day: the 200-meter and 400-meter individual medleys. Meanwhile, Mr. Phelps is eschewing the 100-meter freestyle race that Mr. Spitz won. That brings us to …

Head to head: Mr. Phelps gets a grade of incomplete for two of their six common events: the 100-meter butterfly and the 4×100-meter individual-medley relay. Comparing their times is unfair: Mr. Spitz never had a chance at the generations of swimsuits to come, nor advances in training and other pool technology. (He did try and fail to make a comeback 20 years after his Munich triumphs, but he was in his 40s at the time.)

Winning margin is a better measure of dominance among peers. And in the four events they’ve both completed (200-meter freestyle, 200-meter butterfly and 4×100-meter and 4×200-meter freestyle relays), Mr. Spitz was more dominant at the Olympics. On average, he won those events by 1.36%; Mr. Phelps’s average winning margin was 0.91%.

But against the backdrop of history, or the history each swimmer faced, it looks more like a tie. Mr. Spitz set new world records by an average of 1.02%, compared to the pre-Olympics records. Mr. Phelps’s new marks exceeded the old ones by 1.05%, on average.

This pattern holds if we widen our lens to take in all their events. Mr. Spitz’s average winning margin was 1.47%, compared to 1.09% for Mr. Phelps. But Mr. Phelps’s new records have exceeded the old by 0.88%, compared to 0.87% for Mr. Spitz.

Here’s how I calculated these percentages: Because the races took longer in Mr. Spitz’s day, a victory of a second meant less than it means now. So I divided the winning margin — the runner-up’s time minus the winner’s — and divided by the runner-up’s time, as a measure of how much faster the winner was. Then I averaged that winning margin, and multiplied by 100 to convert to a percentage. Similarly, to compare records, I divided the margin between the new record and the old one by the old record; then averaged the record margin across events and converted to percentages. (Throughout this post I use average to indicate mean; I checked the median, and it made little difference.)

Individual events: The above analysis includes relay events, which depend on three other swimmers. Mr. Spitz swam three of these; Mr. Phelps has swum two and has one more to come. Discard these, and Mr. Spitz has a slight edge in winning margin (1.44% to 1.33%) and record margin (0.58% to 0.5%).

Context: Mr. Phelps is hardly the only record-setter in Beijing; four of the six men’s swimming events he’s skipped that have finished have set new records at the Water Cube. Many writers have attempted to explain why Olympians are swimming so fast this time around. All this attention on records might suggest that Mr. Phelps’s own records are diminished; by this logic, if conditions allow for faster swimming, a new record is a mere byproduct of winning.

That may be so, but it’s worth noting that records were falling at a similar clip in 1972. In the eight events Mr. Spitz skipped, two records were tied and four were broken.

Amid these record-breaking environments, however, both men surpassed their counterparts in other races. As I mentioned earlier, Mr. Phelps’s new records have exceeded the old by 0.88%. In the other six completed swimming races, including those that didn’t feature new records, the winning time was an average of 0.24% better than the pre-Olympics record. Mr. Spitz’s records beat the old ones by an average of 0.87%; his counterparts beat record times by an average of 0.34%.

Verdict: So far, it’s essentially a tie. Each swimmer won and set a world record in each race. Mr. Spitz was slightly more dominant in his races. However, Mr. Phelps’s records have been slightly more impressive, both compared to the old records and compared to other record breakers at the same Olympics.

Mr. Phelps will be acclaimed as the greatest Olympian ever if he wins his last two events. He has a chance to allay all doubts about that status if he sets records in both and wins them in routs.

Update: See how their numbers compare after Mr. Phelps completed his eight events.

More Olympics numbers: The print Journal’s Ian Johnson has a fascinating piece about ranking nations by medal counts: Should total medals count, or just golds? See more Olympics coverage here.

Sources: Olympic.org, NBC Olympics and databaseOlympics.com

My print column this week discusses a study published in the journal Obesity that projects recent increases in the prevalence of obesity among Americans into the future, and arrives at some alarming numbers. Among them: By 2048, 100% of American adults could be overweight. The problem with that logic is that a linear trend can’t continue past 100%. For instance, recent trends in abandonment of landlines suggest that more than 100% of American adults will not have landlines by that same year, but as readers of this blog know, only public figures regularly exceed 100%.

What do you think? Do the obesity projections seem reasonable? If they help galvanize health officials to fight the problem, does that justify fuzzy numbers? How would you project obesity rates to 2048? Please let me know in the comments.

What bundle of goods is needed to participate in society, and in its absence is a marker of poverty? It includes “a self-catering vacation, a cell phone, and enough booze to get drunk twice a month,” as Slate describes new poverty standards based on focus-group input. The findings are questionable, but the Slate article provides a useful overview of why poverty standards could use an overhaul, or at least an update. For more on counting the world’s poor, see this Numbers Guy column from last year.

Since you’re likely to hear a lot in the next few months about how Barack Obama is the most liberal U.S. Senator, a claim based on National Journal’s rankings, it’s worth knowing more about how the rankings work. According to the New Republic, they don’t. It’s hard to account for missed votes, an inevitability for presidential candidates. When Sens. Obama and John McCain do show up, their every vote is their votes may be* categorized as liberal or conservative even though not each issue is so clearly partisan.

Polling has been a recurring topic on this blog during the campaign. In a recent cover story, CQ Weekly analyzed the unique challenges this presidential election poses for pollsters. These issues include the role of race; the increasing importance of independent voters; and chronic problems in polling such as declining response rates.

The FBI estimates the odds of a certain kind of DNA match in criminal investigations being coincidental at 113 billion to one. The Los Angeles Times, however, reports new findings calling into question that estimate. “At stake is the credibility of the compelling odds often cited in DNA cases, which can suggest an all but certain link between a suspect and a crime scene,” the paper reports.

*Correction: This blog post originally and incorrectly stated that National Journal rankings categorize all Senate votes as liberal or conservative.