A Wikipedia entry:
The Heart Sutra, it is generally thought, is likely to have been composed in the 1st century CE in Kushan Empire territory, by a Sarvastivadin or ex-Sarvastivadin monk.
The earliest record of a copy of the sutra is a 200-250 CE Chinese version attributed to the Yuezhi monk Zhi Qian.
It was supposedly translated again by Kumarajiva around 400 CE, although John McRae and Jan Nattier have argued that this translation was created by someone else, much later, based on Kumarajiva's Large Sutra.
Zhi Qian's version, if it ever existed, was lost before the time of Xuanzang, who produced his own version in 649 CE, which closely matches the one attributed to Kumarajiva.
Xuanzang's version is the first record of the title "Heart Sutra" (心經 xīnjīng) being used for the text, and Fukui Fumimasa has argued that xinjing actually means dharani scripture. Sichuan, and subsequently chanted it during times of danger in his journey to the West.
However, based on textual patterns in the Sanskrit and Chinese versions of the Heart Sutra and the Mahaprajnaparamita Sutra, scholar Jan Nattier has suggested that the earliest (shortest) version of the Heart Sutra was probably first composed in China in the Chinese language from a mixture of Indian-derived material and new composition, and that this assemblage was later translated into Sanskrit (or back-translated, in the case of most of the sutra). She argues that the majority of the text was redacted from the Larger Sutra on the Perfection of Wisdom, which had originated with a Sanskrit Indian original, but that the "framing" passages (the introduction and concluding passages) were new compositions in Chinese by a Chinese author, and that the text was intended as a dharani rather than a sutra.
[9][19][20] The Chinese version of the core (i.e. the short version) of the Heart Sutra matches a passage from the Large Sutra almost exactly, character by character; but the corresponding Sanskrit texts, while agreeing in meaning, differ in virtually every word.[21] Furthermore, Nattier argues that there is no evidence (such as a commentary would be) of a Sanskrit version before the 8th century CE[22], and she dates the first evidence (in the form of commentaries by Xuanzang's disciples Kuiji and Wonch'uk, and Dunhuang manuscripts) of Chinese versions to the 7th century CE. She considers attributions to earlier dates "extremely problematic". In any case, the corroborating evidence supports a Chinese version at least a century before a Sanskrit version.[23] This theory has gained support amongst some other prominent scholars of Buddhism, but is by no means universally accepted.[24]