Spinal codes are a new kind of adaptive rateless codes. By analyzing the theory, consisting of hash-function encoding, modulation, maximum-likelihood decoding, and puncturing, the data structure was redesigned, and simulation of the whole transmission process was accomplished. The transfer rate was tested with both additive white Gaussian noise channel (AWGN channel) and binary symmetric channel (BSC). The experiment with AWGN demonstrated the rightness of our design and showed that spinal codes are extremely close to the Shannon Limit. What?s more, validity of the data structure was also proven with BSC. These results show that spinal codes are efficient and practical, which will be very useful in the near future.