Let's talk about crime shows. In a non-stop media stream filled with reality shows, cooking competitions and whatever's happening in Westeros, police procedurals are probably as close as most of us are going to get to seeing science portrayed in prime time. And the techniques that crime-fighters use to catch bad guys vary from show to show, but a lot of the time it involves forensics, which is basically the use of science in the field of law, in this case, criminal law.

Different kinds of forensic investigators have different rules, like analyzing crime scenes or running tests in the lab, and they can specialize beyond that, focusing on analyzing DNA or bullets for example. Generally, they all have an undergraduate degree in a scientific field, like chemistry or biology, or a more targeted degree in forensic science itself. Some have a graduate degree too, and medical examiners, or MEs, usually have a degree in medicine. But they all have one thing in common: using science to find, gather, and analyze evidence that can be used in court.

However, Hollywood seems to think that real science doesn't always make for entertaining TV, so writers tend to take some liberties with how forensics really works. Most of the time they aren't completely off the mark, for example the tests they use on the show might actually exist, but they wouldn't be nearly as fast or accurate in real life. And the technology they use is just ridiculous. We are here to clear that up, and talk about what forensics can actually do, which turns out to be pretty interesting all by itself, and to do that we're going to solve a hypothetical crime. 

So here's our case. Someone finds a dead guy in an alley in Chicago. The cops secure the scene and the forensic investigators show up around 11 pm to gather clues. When they go through the victim's pockets they find a receipt for a bottle of soda from a nearby convenience store time-stamped at 5 pm, 6 hours earlier. And, according to the I.D. in his wallet, his name is Bob.

The medical examiners wanna know how long Bob has been dead which could be key to finding and catching his killer. So there are a few things they can check and they all happen to end in the word "mortis", which makes sense since that just means "death" in Latin.

First, there's livor mortis, or how the blood pools. Now that Bob's heart isn't distributing his blood anymore, it just goes where gravity takes it and that makes the skin look purplish from the outside. But if a body's been dead for more than 12 hours, the blood will have coagulated, or dried. It stays in place, and if you shift the body, the blood won't pool in a new spot. Now Bob's blood seems to still be very liquid, so he's been dead less than 12 hours, though of course the examiners already knew that since he was alive and well in the convenience store only 6 hours ago. 

Next they check if rigor mortis, the stiffening of the muscles after death, has set in, and rigor mortis is proof that your muscles work in kind of the opposite way than you might expect, since running and lifting weights and doing things that require muscles is hard, you might think making your muscles contract requires a lot of energy. But, that's not true. Your body actually uses energy to make your muscles relax, not contract, so after someone dies and their muscles stop getting chemical energy, their muscles can't un-contract so their bodies stiffen. This effect starts about 2 hours after death and lasts until about 36 hours in, when the muscles decompose enough that they can't hold their position anymore. In Bob's case, rigor mortis does seem to have set in. He's frozen in place, so the body's probably more than 2 hours old. 

They'd like to get a more accurate number though. If the body is close to 6 hours old, that means he was probably murdered right after he left the store. So they take the body's temperature (rectally, a detail they don't normally show in crime dramas) and it's 29 degrees Celsius. Now normally, a body loses heat at a rate of about 1.5 degrees Celsius per hour, a process known as algor mortis. When Bob was alive, his body temperature would have been 37 degrees, so its lost 8 degrees so far. You'd think Bob's been dead for very close to 6 hours, and in a TV show, the ME would probably say that.

But there's a problem. This is a cold winter evening in Chicago. It's about 5 degrees outside. The body is going to lose heat a lot faster to the colder air, but it's hard to tell exactly how fast. Given all the information they've gathered, our MEs put the time of death between 5 pm and 7 pm. There's no way to tell if Bob was murdered right after he left the store or two hours later.

So detectives head to the store and ask to review the security camera footage, hoping they'll be able to figure out if anyone was with Bob when he bought his drink. Turns out that as Bob left the store the camera picked up someone quickly emerging from behind a nearby tree to follow him down the street. But it was so far away that the stalker's face is all pixelated and blurry. You can hardly even tell it's a face, let alone whose it is.

Now if this were a TV show, usually the detectives would zoom in on the face and "enhance the image" somehow, and then run the magically clear photo through a facial recognition database. And then maybe the next part of the story is that they get a match, which leads them to another clue. But in real life, there's no way they could enhance a picture like that.

When a camera captures a digital image, it's recorded as data that forms a map of the colors in each point, or pixel, in the picture. And those pixels cover bigger or smaller amounts of space depending on the resolution, or how many pixels there are in that image. The color of each pixel is recorded as the average of all the colors within that space. But once the color is stored as the average, that's it. You can't enhance the resolution of a photo because there's no way to tell which amounts of which colors went into that average in each pixel.

Like let's say your camera has 8 megapixels, which is pretty typical for a smartphone. That means it takes a picture with 8 million pixels in it. That sounds like a lot, but let's just say you want to take a picture of something really small or really far away, like a person at the other end of a field. You can zoom in a lot, but you still probably won't be able to make out much detail. Whatever's written on their t-shirt, for example, might just look like a few blocky, dark green squares. And there's no way you could enhance those squares to see that the dark green pixels are just averaging together the bright green letters on a black background that spell out Sci Show. Which of course are available at DFTBA.com/SciShow. If you wanted to be able to make out what's on the shirt, you'd need a lot more pixels that would each depict a smaller area of that mysterious figure.

But let's say our real-life detectives look through some more of the footage and realize the person who was following our victim was actually back in the store about three hours later. They can tell because he was wearing the same clothes. The camera captures him as he puts something down on a shelf and leaves. They get a close-up picture of his face and run it through the database.

Facial recognition actually has a long history because it's one of those things that humans tend to be very good at, but it's hard to get computers to do well. Humans are excellent at finding patterns, but the computers have to be taught what to look for. Human features, it turns out, are arranged in very specific ways, but the specifics are unique to each person. For example, everyone has a certain curvature to their eye sockets, or distance between their nose and mouth.

Those dimensions are different for everyone, but computers can be programmed to measure them and then use the data to identify faces. Together, these metrics make up a face-print and there are actually databases of face-prints compiled from things like mugshots. A computer can take an image of a face, like the one of the man following Bob, and compare his face to the ones already in the database.

On TV shows, that database will usually be shown as a sophisticated system with all the data in one place. All the detectives have to do is type some commands on a keyboard and the computer starts cross-referencing with every picture ever taken in the country, but that kind of law enforcement database doesn't really exist, at least not yet. The FBI is working on what they say will be the world's biggest database of bio-metrics, the biological traits that can be used as ID, which tens of millions of records, but for now, if cities has searchable face-print databases at all, they're usually local.

Chicago, for example, has one called NeoFace that looks for matches in the police photo database. in our case, the investigators catch a lucky break. When they run the face-print against the Chicago police department's database of mugshots, they find a match. His name is Charlie, he owns a hardware store a few blocks away. When the investigators inspect the shelf they saw him putting something on, they find a wrench with a dark red stain.

Thinking that it might be their murder weapon, they take a swab of whatever's on the wrench, then they use something called a kassel-meyer test to see if it's blood. On TV, you might just see them spraying some liquid onto the swab to see if it changes color, but in real life, they'll need to use two different substances. First, they add the chemical phenolphthalein to the swab, then a couple of drops of hydrogen peroxide. If there's blood in the sample, the two compounds will react with each other, turning the phenolphthalein a vivid shade of pink.

Blood contains hemoglobin, which acts as a catalyst. Basically, the substance that makes a reaction happen. With hemoglobin's help, the peroxide reacts with the hydrogen peroxide in phenolphthalein and becomes water. The new hydrogen-less form of phenolphthalein then turns pink. If there were no blood on the wrench, the reaction wouldn't happen because it wouldn't have a catalyst to help it along, but in this case the swab from the wrench does turn pink meaning that the stain is probably blood, so the investigators take the wrench for further testing.

Back at the lab, they run a DNA analysis on the blood from the wrench and compare it with Bob's. If it's a match, they've probably found their murder weapon, and their murderer. Now this test actually works pretty much like it does on TV. DNA is the molecule that makes you who you are, long strings of four different compounds or base pairs, in a particular order, and everybody has their own unique set, except for identical twins.

So if you have a DNA sample, that's a really good way to identify someone, but forensic teams don't just sequence everyone's DNA, instead they use a technique known as STR analysis to match DNA samples. It's based on the idea that everyone's DNA has certain sections with repeating patterns of base pairs, and the number of times the pattern repeats varies from person to person. The STR looks at 13 of those repeating sections, and the odds of two people having the exact same base pairs in all 13 are about 1 in a billion.

Meaning there are probably only about 6 other people in the entire world who have the same STR profile as you, and forensic experts figure that's accurate enough. Plus, it takes less than an hour and a half to run. So in our case investigators find that the blood on the wrench did come from Bob, which certainly makes Charlie a suspect, but there are still questions.

Was Bob's encounter with the wrench what killed him? Was someone else involved besides Charlie? Unfortunately I can't answer those questions for you because we're out of time and Game of Thrones is about to come on, but thanks for watching, and thanks especially to our patrons on Patreon who make this show possible. If you want to help us make episodes like this, just go to Patreon.com/SciShow, and don't forget to go to YouTube.com/SciShow and subscribe.