1. Punched Cards: How Holes in Paper Cards Represented Text
Long before computers entered homes, data containers were thick paper cards with holes punched in them. These were called punched cards. The idea of using punched paper or cards to teach patterns to looms originated in the 1700s and early 1800s; the Jacquard loom in France used this method in the early 1800s [5]. Herman Hollerith applied this concept to statistical tabulation. In the 1890 U.S. Census, machines were used that passed needles through the holes to complete electrical circuits and count the results [6][7]. IBM’s history page notes that Hollerith got the idea from conductors who recorded passenger features on tickets by punching holes [4].
In 1928, IBM released a card with 80 columns and 10 rows of holes. An improved version in 1930 had 12 rows. The combination of holes in a single column represented one number or character [4][5]. One card held about 80 characters. For a computer program, this was about one line, or roughly 80 bytes according to IBM [4]. The size was approximately 187 × 83 mm, with a thickness of about 0.18 mm, similar to U.S. paper money of the time [5]. Data could be recorded by holes because machines only needed to distinguish whether a hole was present or absent. If a code was agreed upon for which hole position meant which character, the arrangement of holes became text. However, one card held only one line. Long programs were saved as stacks of many cards.
2. Magnetic Tape: How Data Was Recorded in Sequence on One Long Strip
Instead of stacking many cards, people wanted to record data continuously on one long strip. Magnetic tape was used for this. The UNISERVO, which came out in 1951 for the UNIVAC I, is considered the first tape drive for commercial computers. It could record 1.44 million digits on a single reel of tape that was ½ inch wide and 1,500 feet long [3].
In the 1970s, hobbyists who built computers used commercial cassette tapes. The Compact Cassette was announced by Philips in 1963, originally intended for dictation recording [9]. Zeros and ones were recorded by replacing them with a low tone and a high tone. The "Kansas City standard," agreed on at a meeting held by Wayne Green of Byte magazine in November 1975, wrote a 0 as four cycles of a 1200 Hz wave and a 1 as eight cycles of a 2400 Hz wave, at 300 bits per second. Loading an 8-kilobyte BASIC program took about 5 minutes [8]. In Japan, the NEC PC-8001, announced in May 1979, offered cassette tape recorders and mini floppy disks as add-on devices [10]. In some countries, radio stations broadcast computer programs, and listeners recorded them onto cassettes to load them [9]. Data converted to sound could be sent via radio.
Tape had a weakness. Even if the desired data was in the middle, you had to fast-forward to that point. It was similar to reading a scroll from the beginning to reach the end. However, because fast-forwarding was possible, it was faster than reading. Still, it was poor at "going to a specific place immediately."
3. Floppy Disks and Hard Discs: How Rotating Discs Allowed Quick Access to Data
With a disc, you could move the read/write head and rotate the disc to reach any location in a short time. In 1956, IBM released the RAMAC 305. It recorded about 5 million characters on 50 metal discs coated with magnetic material. Unlike tape or punched cards, it allowed immediate access to data [3]. This was the beginning of the hard disk.
In 1971, IBM released the "Minnow," an 8-inch read-only drive for loading microcode (factory-written instructions) into large computers. It used a disc made of Mylar (plastic film) coated with magnetic material, with a capacity of 80 kilobytes [2][11]. The name "floppy" comes from this softness. The 8-inch and 5.25-inch versions had soft outer sleeves and were literally floppy. Sony’s 3.5-inch version, released in 1981, changed to a hard plastic case with a sliding metal shutter [2]. It is explained that this made it harder for fingers to touch the disc and reduced scratches [1]. The inner disc remained soft, but the case became hard, and the name stayed the same. The save button icon depicts this 3.5-inch version.
Here are the numbers. The "1.44 MB" of a high-density 3.5-inch floppy is 1,474,560 bytes [1]. If one punched card is counted as 80 bytes, this is about 18,400 cards. Since one card is about 0.18 mm thick, stacking them would reach about 3.3 meters.
4. LaserDisc and DVD: What Determined the Size of Optical Discs
Optical discs changed the reading tool from magnetic heads to laser light. LaserDisc (LD) was released in the U.S. in December 1978, and Pioneer released the "LD-1000" in Japan in October 1981 [13]. The diameter was 30 cm, about the size of an LP record. The Japanese Wikipedia explains that it is larger than CDs or DVDs because it recorded video in analog form (as wave shapes). The recording length was 30 minutes per side (CAV format) or 60 minutes (CLV format) [13].
Sony released the player "CDP-101" in October 1982, for CDs with a diameter of 12 cm [14]. DVDs were released in Japan in November 1996, also with a diameter of 12 cm. The light wavelength changed from 780 nm for CD to 650 nm for DVD, and the track spacing (distance between recording lines) became finer, from 1.6 μm to 0.74 μm. A single-layer side can record 4.7 GB, about 6 times that of a CD. Video is compressed using MPEG-2, allowing over 133 minutes on one side [15].
Thus, LD used 30 cm for up to 60 minutes of analog video, while DVD used 12 cm for over 2 hours of compressed digital video. Capacity was determined by the recording method, not just size. DVD is 12 cm to match the CD size and utilize existing player designs [15]. However, picture and sound quality differ, so numbers cannot be simply compared.
There is a famous story about why CD is 12 cm. It is said that Sony’s Norio Ohga wanted a size that could fit Beethoven’s Symphony No. 9 on one disc. English Wikipedia notes that Philips’ official history states this. On the other hand, Philips engineer Kees Immink argues that the recording method (EFM) was not yet decided in December 1979 when 12 cm was adopted [14].
5. Flash Memory Cards: How Data Is Recorded Without Moving Parts
All previous containers had something that moved. Cards were fed by machines, tape was wound, and discs rotated. Flash memory records zeros and ones by trapping electrons. It has no moving parts and retains data when power is off [16]. It was created by Toshiba engineers in the 1980s [16].
In 1994, SanDisk released CompactFlash (CF). The Computer History Museum timeline states it was quickly adopted, especially popular in digital cameras, and was durable with large capacity [3]. Its size is 43 × 36 × 3.3 mm. Compared to the PC Card used in laptops at the time (85.6 × 54 × 3.3 mm), the thickness is the same, but the area is about one-third [17]. The terminals were a smaller version of the PC Card's 16-bit standard, and with an adapter, it fit into PC Card slots. There was also a 1-inch hard disk called "Microdrive" of the same size, so both flash and small hard disks fit in the same slot [17].
Usage in cameras varied by type. English Wikipedia writes that in consumer digital cameras around 2001, SmartMedia accounted for about half, while CF was strong in professional cameras [17]. The CompactFlash Association also states that companies making professional cameras continued to use CF [18]. SD cards were created in 1999 by SanDisk, Matsushita (Panasonic), and Toshiba, and the SD Association was formed the following year. The size is 32 × 24 × 2.1 mm, and microSD is 15 × 11 × 1 mm. Capacity categories include SDXC up to 2 TB and SDUC up to 128 TB [19].
6. Why Newer Containers Hold More Data Despite Being Smaller
The reason for "smaller but more" is that the area used by one record became smaller. Even for the same 3.5-inch floppy, capacity grew from about 280 kilobytes around 1983 to 1.44 megabytes, about 5 times [12][1]. For optical discs, track spacing became less than half as fine from CD to DVD (1.6 μm to 0.74 μm) [15]. For flash memory, the information in one cell increased (NEC demonstrated 2 bits per cell in 1998, and Samsung brought 3 bits per cell to products in 2010), and 3D NAND, which stacks chips vertically, was added [16].
This is similar to writing in small letters in a notebook: you can write more on the same page. However, just making letters smaller isn't enough. Thin laser light, thin heads, and circuits to distinguish fine records had to advance together.
Let’s look at numbers. A 128 GB SD card (1 GB = 1 billion bytes) is equivalent to about 1.6 billion punched cards, if one card is 80 bytes. Stacking them at 0.18 mm each would reach about 290 km, which is about 75 times the height of Mount Fuji.
The small capacity of old storage was not only because the containers held little. The data people put into them was also small. A full-width Japanese character is 2 bytes in Shift_JIS [20]. One 400-character manuscript paper holds a maximum of about 800 bytes. One 1.44 MB floppy can hold about 1,800 pages of text. Today, smartphone photos are often several megabytes each and may not fit on one floppy (this varies by model and settings).
7. How to Investigate Data Sizes Yourself Using a Photo on Your Smartphone
Look at the file size of one photo on your smartphone. Many models display megabytes in the photo's "Info" screen. Compare that number to the 1.44 MB of a floppy and calculate how many floppies it would take. How different is a text-only memo from one photo or one video?
For further reading, the Computer History Museum timeline [3] and IBM’s history pages [4][11] are good entry points. IBM’s floppy page has a sentence describing the 8-inch capacity as "equivalent to 3,000 punched cards." This does not match the first 80 kilobytes cited in this notebook (about 1,000 cards). Which period of 8-inch disk is being referred to? Comparing with other sources is one way to investigate.